<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: RippleX Developers</title>
    <description>The latest articles on DEV Community by RippleX Developers (ripplexdev).</description>
    <link>https://dev.to/ripplexdev</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Forganization%2Fprofile_image%2F4275%2Fc928fdb3-3be0-427c-971c-2224f6c6224a.jpeg</url>
      <title>DEV Community: RippleX Developers</title>
      <link>https://dev.to/ripplexdev</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/ripplexdev"/>
    <language>en</language>
    <item>
      <title>Memory in an XRPL Node: Explicit Heap Trimming after Cache Sweeps</title>
      <dc:creator>Valentin Balaschenko</dc:creator>
      <pubDate>Thu, 05 Nov 2026 16:05:46 +0000</pubDate>
      <link>https://dev.to/ripplexdev/memory-in-an-xrpl-node-explicit-heap-trimming-after-cache-sweeps-4bi9</link>
      <guid>https://dev.to/ripplexdev/memory-in-an-xrpl-node-explicit-heap-trimming-after-cache-sweeps-4bi9</guid>
      <description>&lt;p&gt;&lt;strong&gt;For operators&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;If you measure your node's memory, read the xrpld process, its RSS from &lt;code&gt;/proc/&amp;lt;pid&amp;gt;/status&lt;/code&gt; or a process exporter, not the &lt;code&gt;systemctl status&lt;/code&gt; figure, which includes the kernel's page cache.&lt;/li&gt;
&lt;li&gt;If you run &lt;code&gt;xrpld&lt;/code&gt; 3.3.0 or later, your node already trims the heap after every cache sweep. There is nothing to configure.&lt;/li&gt;
&lt;li&gt;If you build or run &lt;code&gt;xrpld&lt;/code&gt; with an allocator other than glibc's, the trim does not apply.&lt;/li&gt;
&lt;li&gt;If your memory graph has become a fine sawtooth since upgrading, that is the trim working, not a fault.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Background
&lt;/h2&gt;

&lt;p&gt;The &lt;a href="https://dev.to/ripplexdev/memory-in-an-xrpl-node-hunting-a-leak-that-wasnt-there-2c3l"&gt;previous analysis&lt;/a&gt; identified the cause of memory growth as allocator behaviour rather than a memory leak. The node's caches released memory as expected at regular intervals, but the process's memory footprint continued to increase. This was due to &lt;a href="https://www.gnu.org/software/libc/" rel="noopener noreferrer"&gt;glibc&lt;/a&gt;'s allocator retaining freed memory instead of returning it to the operating system. The application was aware of the timing of large memory releases, but the allocator was not informed of these events.&lt;/p&gt;

&lt;p&gt;glibc provides the &lt;a href="https://man7.org/linux/man-pages/man3/malloc_trim.3.html" rel="noopener noreferrer"&gt;&lt;code&gt;malloc_trim&lt;/code&gt;&lt;/a&gt; function, which instructs the allocator to return unused memory to the kernel. This post describes the implementation of explicit heap trimming, the criteria for invoking &lt;code&gt;malloc_trim&lt;/code&gt;, the trigger points considered, why allocator-level alternatives were not pursued, associated performance costs, and the observed impact during controlled testing and production deployment on Mainnet validators.&lt;/p&gt;

&lt;p&gt;The path from finding to release took most of a year. Candidate trigger points were tested on a lab validator in November 2025, the change was merged into the development branch in February 2026 and measured in the performance lab, and it then ran on a Mainnet node through the 3.2.0 release-candidate period. It was held back from 3.2.0 as a precaution while a consensus-stability question was investigated, so 3.2.0 as released does not contain it although its release notes initially listed the change. The investigation found no effect from trimming, and the change shipped in 3.3.0 in August 2026.&lt;/p&gt;

&lt;p&gt;The change was introduced in &lt;a href="https://github.com/XRPLF/rippled/releases/tag/3.3.0" rel="noopener noreferrer"&gt;xrpld version 3.3.0&lt;/a&gt; and is now active on all nodes running that release or later. On Ripple's Mainnet validators it lowered steady-state resident memory by 13% to 23%. The following sections detail its operational effects.&lt;/p&gt;

&lt;h2&gt;
  
  
  Allocator retention heuristics and explicit heap trimming
&lt;/h2&gt;

&lt;p&gt;glibc's memory allocator is designed for efficient repeated allocations by placing freed memory onto free lists, which avoids system calls for similar-sized requests. To reduce lock contention, glibc implements multiple independent heaps called arenas. Each active thread is assigned an arena, and on 64-bit systems, the allocator can create up to eight arenas per processor core (see &lt;a href="https://man7.org/linux/man-pages/man3/mallopt.3.html" rel="noopener noreferrer"&gt;mallopt(3)&lt;/a&gt;). In multi-threaded applications, this results in free memory being retained across many arenas simultaneously.&lt;/p&gt;

&lt;p&gt;Left to itself, glibc returns freed memory to the operating system in two scenarios:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Large allocations.&lt;/strong&gt; Allocations at or above the &lt;code&gt;M_MMAP_THRESHOLD&lt;/code&gt; are given their own memory mapping from the kernel with &lt;code&gt;mmap&lt;/code&gt;, and that mapping is removed as soon as they are freed.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Free space at the top of an arena.&lt;/strong&gt; Smaller allocations are handled within arenas. An arena releases memory only if contiguous free space at the top exceeds &lt;code&gt;M_TRIM_THRESHOLD&lt;/code&gt;, which defaults to 128 KiB. Free memory located elsewhere in the arena is retained regardless of its size.
Both thresholds are dynamic. By default, glibc increases the &lt;code&gt;mmap&lt;/code&gt; threshold when a block larger than the current threshold is freed, raising it to the size of that block, up to a maximum of 32 MiB on 64-bit systems. The trim threshold is set to twice the &lt;code&gt;mmap&lt;/code&gt; threshold. As a result, if a process occasionally frees large blocks, the allocator becomes less aggressive in trimming memory. There is no automatic adjustment in the other direction. The kernel cannot signal memory pressure to the allocator, and the allocator does not revisit previously retained memory.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;code&gt;xrpld&lt;/code&gt;'s allocation pattern is poorly handled by these heuristics. The heap is dominated by small objects, such as tree nodes and ledger entries, allocated in large numbers into caches with varying lifetimes. When caches evict, freed chunks are interleaved with live allocations, so the top of each arena is rarely free and automatic trimming is infrequent. Freed memory therefore remains resident across many arenas, and the process grows until the retained memory matches the application's allocation churn, the rate at which it allocates and frees short-lived objects. This retained volume was the growth measured in the previous post.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;malloc_trim&lt;/code&gt; is the allocator's explicit release path. Since glibc 2.8 it iterates over every arena and returns every whole page of free memory to the kernel, wherever that page sits within the heap, using the &lt;code&gt;madvise&lt;/code&gt; system call to release pages, or &lt;code&gt;sbrk&lt;/code&gt; to shrink the top of the main heap (&lt;a href="https://man7.org/linux/man-pages/man3/malloc_trim.3.html" rel="noopener noreferrer"&gt;malloc_trim(3)&lt;/a&gt;). It is not limited to the top of the heap, which is what makes it effective where the automatic path is not. Its single argument, &lt;code&gt;pad&lt;/code&gt;, reserves an amount of free space at the top of the main heap. Only the main heap honours it. The per-thread arenas ignore it.&lt;/p&gt;

&lt;p&gt;There are two main limitations to &lt;code&gt;malloc_trim&lt;/code&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;It does not move live data.&lt;/strong&gt; Any page containing at least one live allocation remains resident, so the memory cost of fragmentation is reduced but not eliminated.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;It is specific to glibc.&lt;/strong&gt; Other allocators, such as jemalloc or tcmalloc, do not implement this function, and instead manage memory release according to their own policies.
Two alternatives were not pursued in this work. Limiting the number of arenas with &lt;code&gt;MALLOC_ARENA_MAX&lt;/code&gt; reduces how many places free memory is retained, at the cost of more lock contention, but it does not return free pages inside an arena, which is where the growth was. Replacing the allocator with &lt;code&gt;jemalloc&lt;/code&gt; or &lt;code&gt;tcmalloc&lt;/code&gt; changes memory behaviour for every build and deployment at once. An explicit trim at a single call site was the least intrusive change, could be evaluated in isolation, and left the allocator untouched for operators who build differently. An evaluation of &lt;code&gt;jemalloc&lt;/code&gt; is under way separately.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;code&gt;malloc_trim&lt;/code&gt; is available and can release memory that the automatic trimming path cannot reach. The primary cost is a scan of the heap and the page faults that occur when released pages are subsequently reused. The remaining question is when to call it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Trigger selection
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;malloc_trim&lt;/code&gt; should be invoked only after the application has released a significant amount of memory. Invoking it at other times pays for a heap scan with little to show for it, and frequent calls compound this cost. The selection criterion was defined as points in &lt;code&gt;xrpld&lt;/code&gt; where the application knows that a large release has just completed. Five candidate locations were instrumented in &lt;a href="https://github.com/XRPLF/rippled/pull/6022" rel="noopener noreferrer"&gt;PR #6022&lt;/a&gt;, with each call tagged by its trigger and the process &lt;a href="https://en.wikipedia.org/wiki/Resident_set_size" rel="noopener noreferrer"&gt;Resident Set Size (RSS)&lt;/a&gt; logged before and after it. The sweep trigger and the two online-delete hooks ran together for 40 hours on a lab validator connected to Mainnet. The order-book and sync-completion triggers were judged by how often they fire, and the per-cache variant was superseded by the application-level call before the run.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Inside each cache sweep.&lt;/strong&gt; The first implementation called &lt;code&gt;malloc_trim&lt;/code&gt; from &lt;a href="https://github.com/XRPLF/rippled/blob/develop/include/xrpl/basics/TaggedCache.h" rel="noopener noreferrer"&gt;&lt;code&gt;TaggedCache::sweep&lt;/code&gt;&lt;/a&gt;, once per cache. Several caches sweep in close succession, so this version needed a rate limiter with a five-second minimum interval and a re-entrancy guard to prevent overlapping calls.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;After the application sweep.&lt;/strong&gt; &lt;code&gt;Application::doSweep&lt;/code&gt; executes all cache sweeps sequentially as a background job, with the interval determined by &lt;code&gt;node_size&lt;/code&gt;. Invoking &lt;code&gt;malloc_trim&lt;/code&gt; once after the final sweep captures all memory releases in the cycle (&lt;a href="https://github.com/XRPLF/rippled/pull/6022/changes/72b34e661549060fac63cb0e52eba09accdb6243" rel="noopener noreferrer"&gt;commit&lt;/a&gt;). This trigger accounted for the majority of memory reclaimed, executed at each sweep interval, and reclaimed memory on nearly every call.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;After online-delete rotation.&lt;/strong&gt; Two hooks in &lt;code&gt;SHAMapStoreImp&lt;/code&gt; were evaluated: one following ledger cache clearance and another after old ledger deletion (same commit). Both triggers were infrequent due to the rarity of rotation events, and most memory released during rotation was already reclaimed by the subsequent sweep.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;After the full order-book rebuild.&lt;/strong&gt; &lt;a href="https://github.com/XRPLF/rippled/blob/3.3.0/src/xrpld/app/ledger/OrderBookDBImpl.cpp" rel="noopener noreferrer"&gt;&lt;code&gt;OrderBookDB::update&lt;/code&gt;&lt;/a&gt; walks the ledger state every 25,600 ledgers, a constant in the code that works out to roughly once a day, and only on nodes with pathfinding enabled (&lt;a href="https://github.com/XRPLF/rippled/pull/6022/changes/ff8b4353bc7473687bbf3d2804916af2e192ab5f" rel="noopener noreferrer"&gt;commit&lt;/a&gt;). It cannot provide regular reclamation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;On sync completion.&lt;/strong&gt; The transition into the &lt;code&gt;full&lt;/code&gt; operating mode in &lt;code&gt;NetworkOPs&lt;/code&gt; (&lt;a href="https://github.com/XRPLF/rippled/pull/6022/changes/2b2b361c87f725077975ee55195302e4ef17c705" rel="noopener noreferrer"&gt;commit&lt;/a&gt;), which happens once after start and then only when the node loses and regains synchronisation. Too rare for regular reclamation, and with a drawback raised in review: a node that has just regained sync is about to begin its heaviest catch-up work, so trimming at that moment means paying to release memory that will be reallocated immediately.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fmupuwr8crdgjzky8nmb0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fmupuwr8crdgjzky8nmb0.png" alt="Same run, calls from the clearPrior trigger" width="800" height="294"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo2qlqsvdshiy2hksw02i.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo2qlqsvdshiy2hksw02i.png" alt="Same run, calls from the clearCaches trigger" width="800" height="294"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7ini8ufc8ue961zfh7te.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7ini8ufc8ue961zfh7te.png" alt="RSS before and after each malloc_trim call from the sweep trigger, over 40 hours" width="800" height="294"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;RSS before and after every &lt;code&gt;malloc_trim&lt;/code&gt; call, by trigger. Lab validator connected to Mainnet, 11–13 November 2025, 40 hours. Green candles are calls after which RSS was lower, red candles calls after which it was higher. Top: the sweep trigger, firing at every sweep interval. Middle and bottom: the two online-delete triggers, firing a few dozen times over the run.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The &lt;a href="https://github.com/XRPLF/rippled/pull/6022#discussion_r2793438904" rel="noopener noreferrer"&gt;decision&lt;/a&gt; documented in the PR was based on two factors: the sweep-based trigger achieved the highest observed memory reclamation, and it operates within a background job context, keeping heap scan overhead off the consensus and RPC paths. Only the sweep-based trigger was retained.&lt;/p&gt;

&lt;p&gt;The best-performing trigger also determined where the call should be made from. Invoking &lt;code&gt;malloc_trim&lt;/code&gt; within each cache's sweep required throttling for safety. A single invocation after &lt;code&gt;Application::doSweep&lt;/code&gt; completes all cache sweeps does not require throttling, as the call frequency is limited by the sweep interval and each invocation addresses the full release cycle. The final implementation includes a single call in &lt;code&gt;Application::doSweep&lt;/code&gt;, with the rate limiter and re-entrancy guard removed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Trim padding
&lt;/h2&gt;

&lt;p&gt;The &lt;code&gt;pad&lt;/code&gt; argument of &lt;code&gt;malloc_trim&lt;/code&gt; reserves free space at the top of the heap so that the next allocations do not have to grow the heap again. A non-zero reserve could in principle reduce the cost of refilling the caches after each sweep. In practice, only the main heap honours &lt;code&gt;pad&lt;/code&gt;, and the per-thread arenas that serve &lt;code&gt;xrpld&lt;/code&gt;'s worker threads ignore it, so a reserve could protect at most the main heap's share of the memory.&lt;/p&gt;

&lt;p&gt;Four 12-hour test runs were executed on the lab validator connected to Mainnet. Trimming was invoked from &lt;code&gt;Application::doSweep&lt;/code&gt; with &lt;code&gt;pad&lt;/code&gt; set to 0, 256 KiB, 1 MiB, and 16 MiB. RSS was sampled at 10-second intervals, and the duration of each trim operation was logged. To eliminate the impact of the initial ramp-up, RSS statistics are reported from the second hour of each run onward.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;code&gt;pad&lt;/code&gt;&lt;/th&gt;
&lt;th&gt;Mean RSS (GiB)&lt;/th&gt;
&lt;th&gt;P95 RSS (GiB)&lt;/th&gt;
&lt;th&gt;Trim calls&lt;/th&gt;
&lt;th&gt;Mean trim (ms)&lt;/th&gt;
&lt;th&gt;P95 trim (ms)&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;15.64&lt;/td&gt;
&lt;td&gt;16.22&lt;/td&gt;
&lt;td&gt;670&lt;/td&gt;
&lt;td&gt;163&lt;/td&gt;
&lt;td&gt;393&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;256 KiB&lt;/td&gt;
&lt;td&gt;15.91&lt;/td&gt;
&lt;td&gt;16.50&lt;/td&gt;
&lt;td&gt;706&lt;/td&gt;
&lt;td&gt;171&lt;/td&gt;
&lt;td&gt;373&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;1 MiB&lt;/td&gt;
&lt;td&gt;16.01&lt;/td&gt;
&lt;td&gt;16.65&lt;/td&gt;
&lt;td&gt;645&lt;/td&gt;
&lt;td&gt;179&lt;/td&gt;
&lt;td&gt;416&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;16 MiB&lt;/td&gt;
&lt;td&gt;15.77&lt;/td&gt;
&lt;td&gt;16.72&lt;/td&gt;
&lt;td&gt;621&lt;/td&gt;
&lt;td&gt;166&lt;/td&gt;
&lt;td&gt;372&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The differences are small. Mean RSS varies by 0.4 GiB across the four settings, under 3%, and mean trim duration by 16 ms, about 10%. Each setting was run once, in sequence, against whatever Mainnet traffic arrived during its twelve hours, and variation between runs from load alone is of the same order. The comparison can therefore rule out a large effect but cannot rank the settings. A small effect is also what the semantics predict, since only the main heap honours the reserve.&lt;/p&gt;

&lt;p&gt;The default was kept. &lt;code&gt;pad&lt;/code&gt; is zero, and it is a compile-time constant rather than a configuration setting, so that the change adds no tuning surface to the node.&lt;/p&gt;

&lt;h2&gt;
  
  
  Memory results
&lt;/h2&gt;

&lt;p&gt;The change was measured in three tests: a lab machine connected to Mainnet, a closed performance network under a fixed transaction load, and two of Ripple's Mainnet validators over the summer.&lt;/p&gt;

&lt;h3&gt;
  
  
  Lab machine
&lt;/h3&gt;

&lt;p&gt;The lab validator connected to Mainnet ran two builds on the same hardware and settings: the unmodified binary, and a binary with the trim triggers as they stood in November 2025, which called &lt;code&gt;malloc_trim&lt;/code&gt; from the application sweep and from the two online-delete points, of which only the sweep call shipped. RSS and Referenced memory were sampled every 10 seconds, and each trim call recorded the process RSS before and after it. The baseline ran for about 24 hours, the trimmed build for about 40, and one run was made of each. The trimmed run is the one shown in the candle charts in the Trigger selection section.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Baseline.&lt;/strong&gt; RSS grew from 14.0 GiB to 32.9 GiB over 24 hours and was still rising when the run ended.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Trimmed.&lt;/strong&gt; RSS climbed to 16 GiB within the first hour, to 17 GiB by the sixth, and then stayed within an 18–20 GiB band for the rest of the 40 hours, drifting upward at about 30 MB per hour.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;At 24 hours of process life,&lt;/strong&gt; RSS was 18.3 GiB in the trimmed build against 32.9 GiB in the baseline, 44% lower. The trimmed value is read from the trim log's per-call RSS record, so the two processes are compared at the same age.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Over the 40 hours,&lt;/strong&gt; the trimmed build made 1,389 calls, 1,173 of them from the sweep. 91% of them reduced RSS, by a median of 75 MB and a maximum of 1.38 GB, and the observed RSS decrease across all calls summed to about 147 GB, against a net RSS growth of about 14 GB.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;RSS minus Referenced&lt;/strong&gt; in the trimmed steady state was about 0.2 GiB on a 19 GiB process, about 1%, so almost all resident memory had been accessed since the window opened.&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Performance network
&lt;/h3&gt;

&lt;p&gt;The performance lab runs a closed five-validator network under a fixed transaction load. Comparison tests with and without &lt;code&gt;malloc_trim&lt;/code&gt; each ran for five hours. Validator nodes used 2.58 GB less RSS with trimming enabled, a reduction of 14.8%. Client-handler nodes used 3.33 GB less, a reduction of 18.2%. The CPU and consensus-timing measurements from the same runs are in the Overheads section.&lt;/p&gt;

&lt;h3&gt;
  
  
  Mainnet validators
&lt;/h3&gt;

&lt;p&gt;Two of Ripple's validators connected to Mainnet, one on the default UNL (dUNL) and one not, ran 3.2.0 without heap trimming from 16 June to 31 July 2026 and 3.3.0 with it from 7 August. The trim call was removed from 3.2.0 the day before its release, so the 3.2.0 builds in this comparison do not contain it, although the release notes initially listed the change. The week between the windows saw the 3.2.1 hotfix release and an unrelated network incident, and was excluded. The 3.2.0 windows close before the incident, and leaving out their final three days changes the plateau means by less than 0.05 GiB. Process RSS is recorded by Prometheus from the &lt;a href="https://github.com/ncabatoff/process-exporter" rel="noopener noreferrer"&gt;process exporter&lt;/a&gt;, and both windows on both nodes open on a fresh start of the process, so the two versions are compared at the same age. The figures below are the plateau each node settled at after its first five days.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Node&lt;/th&gt;
&lt;th&gt;3.2.0, no trim&lt;/th&gt;
&lt;th&gt;3.3.0, trim&lt;/th&gt;
&lt;th&gt;Change&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Non-dUNL validator (&lt;code&gt;online_delete&lt;/code&gt; 4,096)&lt;/td&gt;
&lt;td&gt;20.1 GiB&lt;/td&gt;
&lt;td&gt;17.4 GiB&lt;/td&gt;
&lt;td&gt;−2.7 GiB, −13%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;dUNL validator (&lt;code&gt;online_delete&lt;/code&gt; 50,000)&lt;/td&gt;
&lt;td&gt;19.4 GiB&lt;/td&gt;
&lt;td&gt;15.0 GiB&lt;/td&gt;
&lt;td&gt;−4.4 GiB, −23%&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzwkz0eqfxk08n9xtesn4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzwkz0eqfxk08n9xtesn4.png" alt="Resident memory of two Mainnet validators, 3.2.0 without trim and 3.3.0 with trim, by days since process start" width="799" height="605"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Resident memory of two of Ripple's Mainnet validators, 3.2.0 without heap trimming and 3.3.0 with a trim after every sweep, each window opening on a fresh start and plotted by days since start. Dotted lines are the plateau means after day five. The dUNL node's two-day rhythm is online delete: it rotates its history every 50,000 ledgers, about every two days, and under 3.3.0 each rotation's release leaves RSS instead of staying with the allocator. The non-dUNL node rotates every 4,096 ledgers, every few hours, so its releases are small and its line is smooth.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This is a comparison between versions, and 3.3.0 differs from 3.2.0 in more than the trim call. The reduction is nonetheless attributable to trimming: nothing else in the release targets memory retention, the production graphs show the sawtooth of release and regrowth that trimming produces, and the reduction is in line with the lab and performance-network tests, where &lt;code&gt;malloc_trim&lt;/code&gt; was the only difference between the test runs compared.&lt;/p&gt;

&lt;p&gt;The Mainnet validators test has the most weight of the three, since these are production nodes. The windows are 44 and 38 days long, the savings appear on both nodes, and they are well above the ordinary drift either node shows over a window of that length.&lt;/p&gt;

&lt;h2&gt;
  
  
  Overheads
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Trim latency
&lt;/h3&gt;

&lt;p&gt;The &lt;a href="https://github.com/XRPLF/rippled/blob/3.3.0/src/libxrpl/basics/MallocTrim.cpp" rel="noopener noreferrer"&gt;&lt;code&gt;MallocTrim&lt;/code&gt;&lt;/a&gt; helper was instrumented to measure the cost of each call. With debug logging enabled, the helper records, for every invocation, the process RSS before and after execution and the wall-clock duration.&lt;/p&gt;

&lt;p&gt;On the lab validator connected to Mainnet, a sweep trim took a median of 141 ms and a 95th percentile of 340 ms, at roughly one call per minute, which is a fraction of a per cent of the sweep interval, spent on a background thread. The four 12-hour padding runs, with 620 to 706 sweep trims each, gave the same picture, with mean durations of 163 to 179 ms. The duration of a call grows with the size of the heap and with the amount of memory freed since the previous sweep, so nodes with a different &lt;code&gt;node_size&lt;/code&gt; or &lt;code&gt;online_delete&lt;/code&gt; configuration may see longer calls.&lt;/p&gt;

&lt;p&gt;The trim operation executes as part of the background sweep job, once all cache sweeps have completed and outside any cache lock, running on the sweep thread. In glibc, &lt;code&gt;malloc_trim&lt;/code&gt; acquires each arena's lock sequentially during trimming. Threads attempting allocation or free operations on a locked arena are blocked until the lock is released. The following subsection presents consensus-timing measurements to assess the practical impact of these waits.&lt;/p&gt;

&lt;h3&gt;
  
  
  CPU and consensus timing
&lt;/h3&gt;

&lt;p&gt;In the performance lab, a five-validator network was used to execute comparison tests with and without &lt;code&gt;malloc_trim&lt;/code&gt;, each for five hours under a fixed transaction load. CPU usage on validator nodes increased by 0.7% with trimming enabled, while client-handler nodes showed a 3.4% decrease. Both changes are small. Consensus timing was evaluated using &lt;a href="https://en.wikipedia.org/wiki/Effect_size#Cliff's_delta" rel="noopener noreferrer"&gt;Cliff's delta&lt;/a&gt; with &lt;a href="https://en.wikipedia.org/wiki/False_discovery_rate#Benjamini%E2%80%93Hochberg_procedure" rel="noopener noreferrer"&gt;Benjamini–Hochberg&lt;/a&gt; correction across all metrics. No changes were detected in ledger close interval, its 80th percentile, voting duration, or transaction-set acquisition time. Per-transaction apply, flush, and build times showed statistically significant differences, as expected given the number of ledgers processed, but all effect sizes were below the 0.147 threshold considered negligible by this method. The test was repeated several times with consistent results, and the figures above are from one round.&lt;/p&gt;

&lt;h3&gt;
  
  
  Stability
&lt;/h3&gt;

&lt;p&gt;Two checks were made on whether trimming affects a node's participation in the network.&lt;/p&gt;

&lt;p&gt;The first comparison was made during the 3.2.0 release-candidate period, when a Mainnet node ran candidate builds with the trim call for 22 effective days and a candidate build without it for 4.65 days. Departures from the &lt;code&gt;full&lt;/code&gt; state occurred at rates of 2.41 and 2.15 times per day, with median durations of 3.16 and 3.13 seconds, respectively. Consensus step-outs were observed at 2.23 and 2.15 times per day. None of these differences, nor the difference in daily counts of &lt;code&gt;SHAMapMissingNode&lt;/code&gt; events, was statistically significant. The comparison is limited by the short duration of the window without the call, so only large effects can be excluded.&lt;/p&gt;

&lt;p&gt;The second check was conducted as part of an investigation into slow initial synchronisation on Devnet during the 3.3.0 release candidates, where new nodes required 36 to 59 minutes to reach full synchronisation. Trimming was evaluated as a potential cause. In the slowest of three instrumented runs, trimming accounted for 305 calls totalling 72 seconds during a 59-minute sync, representing about 2% of wall time, with approximately 19 seconds occurring within the synchronisation window. Sync time did not vary with sweep interval settings of 60, 600, or default values, and trim pauses did not align with the acquisition timeouts that defined the slow runs. The investigation found no evidence that trimming contributed to slow sync times. The causes of the slowdown lay elsewhere and are outside the scope of this post.&lt;/p&gt;

&lt;h2&gt;
  
  
  Outstanding questions and next steps
&lt;/h2&gt;

&lt;p&gt;Introducing explicit heap trimming after the cache sweep consistently reduced resident memory usage on &lt;code&gt;xrpld&lt;/code&gt; nodes across all measured environments: by 13% to 23% on Ripple's Mainnet validators, by 15% to 18% in the performance network, and by 44% on the lab machine under the heavier load of late 2025. The call cost a fraction of a per cent of the sweep interval on a background thread, and had no measurable effect on consensus timing or node stability. This change was released in version 3.3.0, implemented at a single call site without additional configuration parameters.&lt;/p&gt;

&lt;p&gt;The observed reduction in resident memory varied widely: 23% and 13% on the two Mainnet validators, 15% to 18% on the performance network, and 44% on the lab machine. This variation exceeds what can be explained by measurement differences alone and requires further investigation. One hypothesis fits the evidence so far: the reclaimed memory is memory the allocator retained after short-lived allocations were freed, so its volume follows the rate of allocation churn, which in turn follows transaction load. The largest reduction was recorded in November 2025, during a period of network-wide memory growth, while the Mainnet measurements were made in the summer of 2026. If transaction volume or composition during that period drove the growth of retained memory, the saving would shrink as load normalised. The stabilisation of untrimmed validators at a plateau is consistent with this hypothesis. Verification will require historical correlation of network load with memory metrics and controlled lab experiments with varied transaction mixes.&lt;/p&gt;

&lt;p&gt;Several areas for further optimisation have been identified:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Trim frequency.&lt;/strong&gt; Performing trimming on every sweep is straightforward but may not be the best schedule. Adjusting the schedule to trim every few sweeps, or only when a sweep releases memory above a defined threshold, could reduce the number of trimming operations while retaining most of the memory savings.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Padding.&lt;/strong&gt; The initial four single runs excluded a large effect from &lt;code&gt;pad&lt;/code&gt; and did not allow for ranking of different settings. Conducting repeated, interleaved runs is necessary to determine whether maintaining a small memory reserve provides measurable benefit.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Residual growth.&lt;/strong&gt; The gradual increase in memory usage observed with trimming enabled reflects the live working set, which cannot be reduced through allocator configuration. Addressing this requires revisiting the cache experiments from part one and re-running them in the updated lab environment with trimming enabled.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Allocation churn at the source.&lt;/strong&gt; Trimming returns memory after the application has freed it. It does nothing to reduce how much the application allocates and frees in the first place, and that churn is what causes the fragmentation the trim works around. The complementary work is to find which code paths allocate the most, using the same profiling as in part one, and to reduce it where possible: allocating fewer short-lived objects, or reusing them instead of recreating them.&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>xrpl</category>
      <category>performance</category>
      <category>linux</category>
      <category>memory</category>
    </item>
    <item>
      <title>Memory in an XRPL Node: Hunting a Leak That Wasn't There</title>
      <dc:creator>Valentin Balaschenko</dc:creator>
      <pubDate>Thu, 05 Nov 2026 15:13:41 +0000</pubDate>
      <link>https://dev.to/ripplexdev/memory-in-an-xrpl-node-hunting-a-leak-that-wasnt-there-2c3l</link>
      <guid>https://dev.to/ripplexdev/memory-in-an-xrpl-node-hunting-a-leak-that-wasnt-there-2c3l</guid>
      <description>&lt;p&gt;&lt;strong&gt;For operators&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;If you measure your node's memory, read the xrpld process, its RSS from &lt;code&gt;/proc/&amp;lt;pid&amp;gt;/status&lt;/code&gt; or a process exporter, not the &lt;code&gt;systemctl status&lt;/code&gt; figure, which includes the kernel's page cache.&lt;/li&gt;
&lt;li&gt;If you run &lt;code&gt;xrpld&lt;/code&gt; 3.3.0 or later, your node already trims the heap after every cache sweep. There is nothing to configure.&lt;/li&gt;
&lt;li&gt;If you build or run &lt;code&gt;xrpld&lt;/code&gt; with an allocator other than glibc's, the trim does not apply.&lt;/li&gt;
&lt;li&gt;If your memory graph has become a fine sawtooth since upgrading, that is the trim working, not a fault.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Memory ballooning
&lt;/h2&gt;

&lt;p&gt;Memory growth in XRPL nodes was observed in late summer 2025. Operators reported increasing memory usage on their validators that did not decrease, including on nodes configured identically to others that were unaffected, and across different hardware platforms. The cause turned out not to be a memory leak but the memory allocator retaining memory the application had already freed, and a fix shipped in xrpld 3.3.0. This post describes how that was established.&lt;/p&gt;

&lt;p&gt;A memory regression identified in one release during that period was addressed within weeks, but this had no observable impact for operators. The observed memory behaviour existed before the regression and persisted after the fix. The underlying issue remained unresolved at the time. Operators monitoring node caches asked why memory did not come back after a cache sweep, the routine pass that clears expired entries.&lt;/p&gt;

&lt;p&gt;Early in the community discussion, it became clear that consistent measurement was essential. Some reports referenced memory usage from &lt;code&gt;systemctl status&lt;/code&gt;, which reflects the memory of the service's &lt;a href="https://docs.kernel.org/admin-guide/cgroup-v2.html#memory" rel="noopener noreferrer"&gt;cgroup&lt;/a&gt;. A cgroup is the kernel's resource-accounting container that &lt;a href="https://www.freedesktop.org/software/systemd/man/latest/systemd.resource-control.html" rel="noopener noreferrer"&gt;systemd&lt;/a&gt; wraps around every service. In contrast, per-process readings from &lt;code&gt;/proc/&amp;lt;pid&amp;gt;/status&lt;/code&gt; or &lt;code&gt;smaps&lt;/code&gt; report only the process memory. These two metrics often differed by several gigabytes due to the kernel's page cache. When &lt;code&gt;xrpld&lt;/code&gt; (formerly &lt;code&gt;rippled&lt;/code&gt;), the XRPL node software, reads from or writes to its database files, the kernel caches these file pages in memory. This cached memory is not included in the process's &lt;a href="https://en.wikipedia.org/wiki/Resident_set_size" rel="noopener noreferrer"&gt;Resident Set Size (RSS)&lt;/a&gt;, the kernel's count of pages held in RAM by the process, because &lt;code&gt;xrpld&lt;/code&gt;'s database library, &lt;a href="https://github.com/CPPAlliance/NuDB" rel="noopener noreferrer"&gt;NuDB&lt;/a&gt;, uses standard file I/O rather than memory mapping. The kernel charges this cached memory to the service's cgroup because &lt;code&gt;xrpld&lt;/code&gt; initiated the I/O. On busy nodes, the page cache can reach several gigabytes, increasing as the node accesses its database and decreasing when the kernel reclaims memory or when &lt;a href="https://xrpl.org/docs/infrastructure/configuration/data-retention/online-deletion" rel="noopener noreferrer"&gt;online delete&lt;/a&gt; removes old database files. These fluctuations can occur within minutes and are not controlled by the application. As a result, the same machine can report very different memory usage figures depending on the measurement method, leading to differing interpretations by operators.&lt;/p&gt;

&lt;p&gt;On a dedicated &lt;code&gt;xrpld&lt;/code&gt; machine, it may appear that the distinction between process memory and total memory usage is irrelevant, since it is all the node's memory in the end. However, the two types of memory are managed differently. The kernel controls the page cache, which grows to use available RAM and causes the system-wide memory figure to reflect total machine capacity rather than the process state. In contrast, the process's own memory cannot be dropped by the kernel, only swapped, and it is the only part that code changes can affect. Per-process memory is therefore the right measure of the application's own behaviour, and all subsequent analysis uses it. &lt;em&gt;RSS&lt;/em&gt; is the coarsest of these, since it counts every resident page whether or not the process still touches it. The next section adds a finer one, the working set, which counts only the memory the process has actually accessed over a period.&lt;/p&gt;

&lt;p&gt;A memory leak was the first suspicion. Agreeing on how to measure memory made it possible to test that suspicion against the alternatives: normal cache operation, a feature retaining more state than expected, or an issue below the application layer. Without the ability to answer this reliably, for any &lt;code&gt;xrpld&lt;/code&gt; version and whatever mix and volume of traffic the node handles, using trusted metrics, any conclusions would be speculative.&lt;/p&gt;

&lt;p&gt;A toolset was assembled to collect and analyse the relevant measurements. The following analysis describes the investigation, the findings, the attempted fixes, and the eventual realisation that the root cause was located in a different layer of the stack.&lt;/p&gt;

&lt;h2&gt;
  
  
  The toolset
&lt;/h2&gt;

&lt;p&gt;The investigation focused on three key questions, with the requirement that no conclusion would rest on a single tool.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;What are the primary sources of the increased memory usage?&lt;/li&gt;
&lt;li&gt;What mechanisms are responsible for the observed memory growth on some nodes but not others? For instance, which operational flows, features, or traffic patterns trigger it?&lt;/li&gt;
&lt;li&gt;How reliable are the measurements under real-world conditions? This had to be answered before the other two meant anything.
The previous section established that the focus is on process memory, not system-wide memory. The remaining concern is the reliability of each measurement tool. Each instrument has limitations: &lt;em&gt;RSS&lt;/em&gt; can be affected by &lt;a href="https://www.gnu.org/software/libc/" rel="noopener noreferrer"&gt;glibc&lt;/a&gt; allocator behaviour, flame graphs may attribute bytes to generic frames if stack traces are incomplete, and the kernel's &lt;em&gt;Referenced&lt;/em&gt; bit provides only an approximation. Using multiple tools and cross-verifying results is necessary to ensure accuracy.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The approach uses three measurement methods, with visualisation applied to the collected data.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A &lt;a href="https://man7.org/linux/man-pages/man5/proc.5.html" rel="noopener noreferrer"&gt;procfs&lt;/a&gt;-based working-set tool, specifically &lt;a href="https://github.com/brendangregg/wss" rel="noopener noreferrer"&gt;Brendan Gregg's &lt;code&gt;wss.pl&lt;/code&gt;&lt;/a&gt;, uses &lt;code&gt;/proc/&amp;lt;pid&amp;gt;/clear_refs&lt;/code&gt; and &lt;code&gt;/proc/&amp;lt;pid&amp;gt;/smaps&lt;/code&gt; to estimate Working Set Size (WSS) over a defined window. This tool is integrated with a custom sampling loop that timestamps each reading and records gross totals. It is a low-overhead, always-on, coarse-grained measurement.&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://ebpf.io" rel="noopener noreferrer"&gt;eBPF&lt;/a&gt;/&lt;a href="https://github.com/iovisor/bcc" rel="noopener noreferrer"&gt;BCC&lt;/a&gt; &lt;a href="https://github.com/iovisor/bcc/blob/master/tools/memleak.py" rel="noopener noreferrer"&gt;&lt;code&gt;memleak&lt;/code&gt;&lt;/a&gt; traces allocator calls using uprobes, which are dynamic tracepoints on user-space functions, with aggregation performed in the kernel. This method is lightweight and provides call stacks for outstanding allocations.&lt;/li&gt;
&lt;li&gt;A custom glibc shim, implemented as an &lt;code&gt;LD_PRELOAD&lt;/code&gt; library, is inserted by the dynamic loader ahead of libc to intercept every allocator call. It interposes on &lt;code&gt;malloc&lt;/code&gt;, &lt;code&gt;free&lt;/code&gt;, &lt;code&gt;calloc&lt;/code&gt;, &lt;code&gt;realloc&lt;/code&gt;, &lt;code&gt;posix_memalign&lt;/code&gt;, and related functions to capture each allocation with a backtrace. Unlike the sampling tools, it records every allocation and every free, so it can answer questions about individual allocations that aggregated data cannot. The cost is a several-fold slowdown and a dedicated run per question, so it serves as the ground truth rather than as a monitoring tool.&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://www.brendangregg.com/flamegraphs.html" rel="noopener noreferrer"&gt;Flame graphs&lt;/a&gt; are generated using Brendan Gregg's &lt;a href="https://github.com/brendangregg/FlameGraph" rel="noopener noreferrer"&gt;FlameGraph&lt;/a&gt; toolkit, with input from the eBPF data, to visualise and analyse call-stack data.
These tools represent a trade-off between overhead and detail. The investigation begins with coarse, low-overhead measurements, and more detailed analysis is performed only when initial results indicate where to focus.&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Layer 1: procfs-based WSS
&lt;/h3&gt;

&lt;p&gt;The primary objective is to determine whether the additional resident memory is memory the process keeps accessing, or memory it holds but has not touched for some time, such as cache entries that were filled once and never read again. &lt;em&gt;RSS&lt;/em&gt; alone cannot make this distinction, as it counts all pages in RAM regardless of recent access.&lt;/p&gt;

&lt;p&gt;The Linux kernel maintains a per-page &lt;em&gt;Referenced&lt;/em&gt; bit, which is set when a page is accessed and can be cleared on demand. &lt;code&gt;wss.pl&lt;/code&gt; uses this mechanism by clearing all bits, waiting for a specified interval, and then measuring how much of the process's memory has been accessed during that period, representing the working set for the window. In cumulative mode (&lt;code&gt;-C&lt;/code&gt;), the tool clears the bits once and observes &lt;em&gt;Referenced&lt;/em&gt; increasing toward &lt;em&gt;RSS&lt;/em&gt; as pages are accessed. Only the clearing operation is costly, so sampling frequency can be adjusted as needed.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6vr0c5kwz2p5p7559eep.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F6vr0c5kwz2p5p7559eep.png" alt="RSS vs Referenced over 11.5 hours, cumulative working-set view" width="800" height="434"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Eleven and a half hours of the ballooning on our System76 lab machine, September 2025 — &lt;code&gt;wss.pl&lt;/code&gt; in cumulative mode, 60 s samples. RSS climbs at ~0.74 GiB/h — spikes partially recover, but the floor only rises. Referenced converges toward RSS — the growth is being accessed, not sitting cold — and the band between them narrows from ~9 GiB to ~3.8 GiB.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The key metric is the difference between &lt;em&gt;RSS&lt;/em&gt; and &lt;em&gt;Referenced&lt;/em&gt;, representing resident memory that has not been accessed since the start of the measurement. This includes caches, fragmented allocator heaps, and retained but idle structures. The gap narrows over the run as &lt;em&gt;Referenced&lt;/em&gt; approaches &lt;em&gt;RSS&lt;/em&gt;, indicating that most held memory has been accessed since the start of the measurement. However, the &lt;em&gt;Referenced&lt;/em&gt; bit does not distinguish between untouched pages containing live but inactive data and other types of memory. Accurately identifying the nature of this gap is central to the investigation.&lt;/p&gt;

&lt;p&gt;Leaks associated with growing live structures, such as caches, queues, or in-flight buffers, result in &lt;em&gt;Referenced&lt;/em&gt; increasing in parallel with &lt;em&gt;RSS&lt;/em&gt;, since the leaked memory continues to be accessed. Cold leaks, where memory is allocated but the pointer is lost and never accessed again, appear only as an expanding gap and cannot be distinguished from other untouched memory using this method. This limitation is one reason WSS is not used in isolation.&lt;/p&gt;

&lt;p&gt;Two caveats:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;The &lt;code&gt;clear_refs&lt;/code&gt; mechanism provides only an approximate measurement. The kernel traverses page tables to set and read the &lt;em&gt;Referenced&lt;/em&gt; bits, introducing a small amount of latency during sampling, around 5% according to Gregg's documentation. Excessive sampling can interfere with the kernel's page-aging information and affect memory reclamation decisions. To mitigate these effects, long sampling intervals are used, and measurements are conducted primarily in a controlled environment.&lt;/li&gt;
&lt;li&gt;Tracking memory utilisation using this method is considered experimental and is documented as such in sources including &lt;a href="https://lwn.net/Articles/642202/" rel="noopener noreferrer"&gt;LWN&lt;/a&gt; and &lt;a href="https://www.brendangregg.com/blog/2018-01-17/measure-working-set-size.html" rel="noopener noreferrer"&gt;Brendan Gregg's WSS post&lt;/a&gt;. This approach provides a baseline measurement rather than a definitive conclusion.&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Layer 2: eBPF + BCC memleak
&lt;/h3&gt;

&lt;p&gt;WSS indicates that memory usage is increasing but does not identify the specific code locations responsible. To obtain this information, BCC's &lt;code&gt;memleak&lt;/code&gt; tool is used.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;memleak&lt;/code&gt; attaches uprobes to the allocator entry points in libc for a specific process. Each allocation is recorded with its size and stack trace, and each free operation removes the corresponding record. After a period of execution, &lt;code&gt;memleak&lt;/code&gt; reports the call stacks associated with the largest outstanding allocations. By sampling these reports at regular intervals, it is possible to track how much memory each subsystem has allocated and not yet freed over time.&lt;/p&gt;

&lt;p&gt;This approach is feasible for production workloads because BCC aggregates data within the kernel, transmitting only summary information to user space. Streaming every malloc event instead would produce more data than could be collected without slowing a process that allocates as frequently as an XRPL node. Testing showed no measurable impact on node throughput. For accurate stack traces, builds must include debug symbols and frame pointers. Without them, call stacks may be incomplete.&lt;/p&gt;

&lt;p&gt;For analysis beyond the largest allocation sites, the byte-attributed data is processed with &lt;a href="https://github.com/brendangregg/FlameGraph/blob/master/flamegraph.pl" rel="noopener noreferrer"&gt;&lt;code&gt;flamegraph.pl&lt;/code&gt;&lt;/a&gt;. The width of each box represents the number of bytes attributed to that stack, with the largest boxes indicating the heaviest allocation sites. This visualisation enables rapid identification of the heaviest allocation sites.&lt;/p&gt;

&lt;h3&gt;
  
  
  Layer 3: the glibc shim
&lt;/h3&gt;

&lt;p&gt;eBPF aggregates allocations by call stack, which identifies the subsystem responsible but not the individual allocation, its size, or the order in which allocations and frees occurred. The glibc shim records exactly that, every allocation and free with a backtrace, at the cost of slowing allocation-intensive code several-fold. It is therefore reserved for targeted runs that answer a specific question about individual allocations. The remainder of this analysis did not need that level of detail, so only the first two layers are used.&lt;/p&gt;

&lt;h2&gt;
  
  
  Following the bytes: TaggedCache
&lt;/h2&gt;

&lt;p&gt;The &lt;a href="https://github.com/XRPLF/rippled/blob/develop/include/xrpl/shamap/README.md" rel="noopener noreferrer"&gt;SHAMap&lt;/a&gt;, which maintains the ledger state, does not manage the lifetimes of its nodes directly. This responsibility is handled by the &lt;a href="https://github.com/XRPLF/rippled/blob/develop/include/xrpl/basics/TaggedCache.h" rel="noopener noreferrer"&gt;TaggedCache&lt;/a&gt;, a partitioned, expiry-based cache that supports the &lt;a href="https://github.com/XRPLF/rippled/blob/develop/include/xrpl/shamap/TreeNodeCache.h" rel="noopener noreferrer"&gt;TreeNodeCache&lt;/a&gt; (for tree nodes), the &lt;a href="https://github.com/XRPLF/rippled/blob/develop/include/xrpl/shamap/FullBelowCache.h" rel="noopener noreferrer"&gt;FullBelowCache&lt;/a&gt; (tracking locally complete subtrees), and several additional structures. Profiling a live node with &lt;code&gt;memleak&lt;/code&gt; identified the TaggedCache subsystem as the primary consumer of outstanding memory. Key components include &lt;code&gt;canonicalize&lt;/code&gt;, which consolidates node copies into a single shared instance, the refresh loop (&lt;code&gt;freshenCaches&lt;/code&gt;) that copies cached entries to the new database file during online delete, and sweep threads responsible for evicting expired entries. Most of the remaining memory usage was attributed to the creation of SHAMap nodes and items managed by the cache, including fetch-and-deserialise operations and inbound ledger-data handlers. The node was configured with &lt;a href="https://xrpl.org/docs/infrastructure/installation/capacity-planning" rel="noopener noreferrer"&gt;&lt;code&gt;node_size&lt;/code&gt;&lt;/a&gt; set to &lt;code&gt;medium&lt;/code&gt;, which determines cache sizes in &lt;code&gt;xrpld&lt;/code&gt;, and online delete was enabled.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fru9wf14sd375yjkl101p.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fru9wf14sd375yjkl101p.png" alt="Flame graph: outstanding heap bytes at peak, cache-related stacks highlighted" width="800" height="435"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Peak outstanding allocations were measured on a test-lab validator connected to Mainnet in September 2025. Here, 'outstanding' refers to memory allocated after &lt;code&gt;memleak&lt;/code&gt; was attached and not yet freed, representing allocations since tracer attachment rather than the process's total memory usage. The graph displays the total outstanding memory as its full width, with each category's share indicated by horizontal width: dark for TaggedCache machinery, light for cache-managed objects, and grey for other allocations. Stack traces were recoloured by category for clarity. The original &lt;a href="https://github.com/XRPLF/rippled/pull/6022#issuecomment-6659691076" rel="noopener noreferrer"&gt;flame graph&lt;/a&gt; is available for reference.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The data was collected with &lt;code&gt;node_size&lt;/code&gt; set to &lt;code&gt;medium&lt;/code&gt;, indicating that the observed memory usage was not due to unusually large cache configurations. This pattern persisted across varying traffic levels. In all peak snapshots of outstanding memory, 84–99.7% was attributed to cache operations, primarily under &lt;code&gt;SHAMapStoreImp::freshenCaches&lt;/code&gt;, which re-populates the new database file from cache keys during online delete. Transient memory spikes were also associated with cache maintenance, proportional to the cache's key population.&lt;/p&gt;

&lt;p&gt;A flame graph attributes memory usage to the code responsible for allocation, not to the structure that ultimately holds the memory. In this codebase, allocation occurs during fetch and deserialisation, and nodes are subsequently managed by the cache through &lt;code&gt;canonicalize&lt;/code&gt;. Therefore, the data identifies a likely source of memory usage, but confirming this would require further experimental validation.&lt;/p&gt;

&lt;h2&gt;
  
  
  Cache experiments
&lt;/h2&gt;

&lt;p&gt;At this stage a likely cause had been identified. Over several months, a series of experiments was conducted to make the cache smaller, cheaper, or better behaved. Each experiment tested a specific theory regarding memory usage.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Theory: some caches may be unnecessary.&lt;/strong&gt; The initial approach was to disable the TaggedCache and observe the impact (&lt;a href="https://github.com/XRPLF/rippled/pull/5439" rel="noopener noreferrer"&gt;#5439&lt;/a&gt;, with initial commits titled "TEST: Disable tagged cache to measure performance"). Disabling the TaggedCache revealed its role as both a cache and the canonicalization point, as well as its function in warming the new database file during online-delete rotation. Complete removal was not feasible. The cache could only be made smaller or shorter-lived, and because rotation warms the new database file from the cache's keys, shrinking it trades memory for post-rotation latency. The same pull request then focused on the NodeObject cache in &lt;code&gt;DatabaseNodeImp&lt;/code&gt;, which stored recently read database records in memory despite the underlying database engine and kernel page cache already retaining these records. Removing this cache reduced memory usage by 3–5% with a minor increase in CPU use.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Theory: the cache structure leads to memory waste on large hardware.&lt;/strong&gt; The TaggedCache is partitioned per hardware thread, causing fixed overhead to scale with the number of cores rather than the workload. Reducing the partition count by half (&lt;a href="https://github.com/XRPLF/rippled/pull/5442" rel="noopener noreferrer"&gt;#5442&lt;/a&gt;) resulted in a 5–6% memory reduction.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Theory: cache entries persist longer than necessary.&lt;/strong&gt; The sweep evicts entries that are unreferenced or expired. To test this, a probe build (&lt;a href="https://github.com/XRPLF/rippled/pull/5910" rel="noopener noreferrer"&gt;#5910&lt;/a&gt;) set all entries to expire immediately, converting each sweep into a full eviction. This approach reduced memory usage by 8–10%.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Theory: some structures are never cleaned.&lt;/strong&gt; The &lt;code&gt;LedgerHistory::ledgersByIndex_&lt;/code&gt; map, which associates ledger sequence numbers with hashes, was not subject to sweeping and accumulated entries indefinitely. This resulted in a small but unbounded growth, a logical leak: entries were added for every validated ledger and never removed (&lt;a href="https://github.com/XRPLF/rippled/pull/5461" rel="noopener noreferrer"&gt;#5461&lt;/a&gt;). It was the only such leak the investigation found. The fix in &lt;a href="https://github.com/XRPLF/rippled/pull/5874" rel="noopener noreferrer"&gt;#5874&lt;/a&gt; bounds the map on nodes with online delete enabled. Full-history nodes, which keep every ledger, still accumulate about 1.4 MB per day. The growth was measured in kilobytes per hour while the overall issue was measured in gigabytes, so it did not account for the observed memory growth.&lt;/p&gt;

&lt;p&gt;Individually, each change showed progress. However, the combined effect was less than expected. While the sum of individual improvements suggested a 20% reduction, the cumulative measurement in a single build was only 5–7%. This discrepancy indicated either unreliable measurements or a lower bound imposed by factors outside the application. Both explanations turned out to be true, and the second was the larger problem. The experiments remain relevant, each demonstrating measurable effects with tools that have since been improved. Further testing in a rebuilt lab environment is planned, but the focus has shifted to addressing the underlying constraint.&lt;/p&gt;

&lt;p&gt;Of the four changes, the NodeObject cache removal (&lt;a href="https://github.com/XRPLF/rippled/pull/5439" rel="noopener noreferrer"&gt;#5439&lt;/a&gt;) was merged and later reverted as a precaution, without any problem having been attributed to it, the partition-count reduction (&lt;a href="https://github.com/XRPLF/rippled/pull/5442" rel="noopener noreferrer"&gt;#5442&lt;/a&gt;) remains open, the sweep-everything build (&lt;a href="https://github.com/XRPLF/rippled/pull/5910" rel="noopener noreferrer"&gt;#5910&lt;/a&gt;) was a probe never intended to ship, and the &lt;code&gt;ledgersByIndex_&lt;/code&gt; fix (&lt;a href="https://github.com/XRPLF/rippled/pull/5874" rel="noopener noreferrer"&gt;#5874&lt;/a&gt;) was merged in October 2026.&lt;/p&gt;

&lt;h2&gt;
  
  
  Released memory is not reclaimed
&lt;/h2&gt;

&lt;p&gt;In October 2025, the investigation moved from periodic snapshots to continuous monitoring. The &lt;code&gt;memleak&lt;/code&gt;-based sampler measured TaggedCache's outstanding bytes, everything its stacks had allocated and not yet freed, every few tens of seconds during hour-long runs. In parallel, WSS tracked &lt;em&gt;RSS&lt;/em&gt; and &lt;em&gt;Referenced&lt;/em&gt;.&lt;/p&gt;

&lt;p&gt;The results were clear. The outstanding allocations from the cache subsystem consistently returned to near zero on each sweep cycle, which occurred approximately every two minutes. The sweep-everything probe from &lt;a href="https://github.com/XRPLF/rippled/pull/5910" rel="noopener noreferrer"&gt;#5910&lt;/a&gt;, which evicts the entire cache at each sweep, amplified this effect. Allocation and deallocation rates matched in real time on the unmodified build. Despite this, &lt;em&gt;RSS&lt;/em&gt; and &lt;em&gt;Referenced&lt;/em&gt; memory continued to increase.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd0qu6oe7ugrktc824ls6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fd0qu6oe7ugrktc824ls6.png" alt="TaggedCache outstanding bytes vs RSS/Referenced, 72 minutes, develop build" width="800" height="587"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Seventy-two minutes on the unmodified develop build, connected to Mainnet, 21 October 2025. Top: bytes allocated by TaggedCache-machinery stacks and not yet freed — the &lt;code&gt;memleak&lt;/code&gt; sampler, ~25 s snapshots. &lt;code&gt;memleak&lt;/code&gt; reports the ten largest outstanding stacks per snapshot, so TaggedCache absent from the top ten plots as zero. Peak 64 MiB, median 2 MiB, every excursion collapsing within minutes (sweep interval ≈ 121 s). Bottom: the same process on the same clock — RSS climbs 13.7 → 15.5 GiB (+1.7 GiB, ~1.5 GiB/h) while Referenced converges toward it (&lt;code&gt;wss.pl&lt;/code&gt;, cumulative mode). That rate is faster than September's 0.74 GiB/h — it varied with the run and the node's age — but the direction never did. One of four captures that day across both builds — the cache's outstanding bytes stayed in the MiB range in every one of them. This run is shown because it has a full RSS/Referenced companion on the same clock.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;This combination of results rules out an application-level cause for the growth, including the cache. The cache's outstanding allocations are measured in MiB, while the observed memory growth is in GiB. Across all stacks reported by &lt;code&gt;memleak&lt;/code&gt;, the median outstanding memory was 0.02 GiB, increasing by approximately 0.1 GiB during the run, compared to 1.7 GiB of &lt;em&gt;RSS&lt;/em&gt; growth. Anything accumulating toward that figure would have appeared in the top ten stacks. Memory that the application has freed but the kernel still counts as resident, and that is still being touched, is not a leak and does not indicate cache malfunction. This behaviour is the allocator's.&lt;/p&gt;

&lt;p&gt;When &lt;code&gt;xrpld&lt;/code&gt; calls &lt;code&gt;free&lt;/code&gt;, glibc's malloc typically retains the pages on its internal free lists, anticipating future reuse by the application. The kernel cannot request memory back from the allocator. Under memory pressure it can drop page cache or swap pages out, but it cannot instruct malloc to release its holdings. glibc only returns free space at the end of each heap when the unused tail exceeds a threshold. Freed memory blocks that are interleaved with live allocations, as is common with caches containing objects of mixed lifetimes, remain stranded within the heap. This is fragmentation. The allocation pattern of an XRPL node, which involves bursts of millions of small, short-lived tree nodes at sweep and rotation boundaries, and the use of per-thread arenas (each busy thread receives its own &lt;a href="https://sourceware.org/glibc/wiki/MallocInternals" rel="noopener noreferrer"&gt;heap region&lt;/a&gt;), leads to fragmentation in every arena. This accumulates to gigabytes of memory that are free from the application's perspective but remain resident according to system metrics.&lt;/p&gt;

&lt;p&gt;The key distinction is between memory &lt;strong&gt;released&lt;/strong&gt; by the application and memory &lt;strong&gt;reclaimed&lt;/strong&gt; by the allocator. The application can release memory, but only the allocator can reclaim it. There was no mechanism in the node's code path to connect these actions.&lt;/p&gt;

&lt;p&gt;The investigation focused on identifying a memory leak. No leak of any consequence was found. The allocator was retaining memory as part of a general-purpose strategy that is poorly suited to this workload. The application can precisely identify when memory is no longer needed, but lacks a mechanism to inform the allocator.&lt;/p&gt;

&lt;p&gt;There is a solution available. glibc provides &lt;a href="https://man7.org/linux/man-pages/man3/malloc_trim.3.html" rel="noopener noreferrer"&gt;&lt;code&gt;malloc_trim&lt;/code&gt;&lt;/a&gt;, which can be invoked by &lt;code&gt;xrpld&lt;/code&gt; at an appropriate point in its operation. The implementation details, performance impact, and effects on memory usage in Mainnet validators will be discussed in the next post.&lt;/p&gt;

&lt;p&gt;The diagnostic toolset fulfilled its purpose by identifying the root cause: the allocator handled &lt;code&gt;xrpld&lt;/code&gt;'s allocation pattern poorly, and neither side had a way to tell the other what it knew.&lt;/p&gt;

&lt;h2&gt;
  
  
  Acknowledgements
&lt;/h2&gt;

&lt;p&gt;This investigation began with reports from operators running the software on Mainnet, who measured the growth on their own nodes, compared notes on how to measure it, and kept asking the questions that turned out to matter. Thanks are due to them, to those who reviewed the change in the open-source repository, and to those who reviewed these two posts.&lt;/p&gt;

</description>
      <category>xrpl</category>
      <category>performance</category>
      <category>ebpf</category>
      <category>linux</category>
    </item>
    <item>
      <title>Lending Protocol V1.1 Performance Test Report</title>
      <dc:creator>Qi Zhao</dc:creator>
      <pubDate>Fri, 09 Oct 2026 17:38:27 +0000</pubDate>
      <link>https://dev.to/ripplexdev/lending-protocol-v11-performance-test-report-3fk5</link>
      <guid>https://dev.to/ripplexdev/lending-protocol-v11-performance-test-report-3fk5</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;Our earlier &lt;a href="https://dev.to/ripplexdev/lending-protocol-performance-test-report-3cjn"&gt;Single Asset Vault &amp;amp; Lending Protocol Performance Test Report&lt;/a&gt; assessed the original Lending Protocol and Single Asset Vault amendments released in xrpld 3.1.0. This report covers the next version of the protocol.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;LendingProtocolV1_1&lt;/code&gt; (&lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0066-lending-protocol" rel="noopener noreferrer"&gt;XLS-66&lt;/a&gt;, &lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0065-single-asset-vault" rel="noopener noreferrer"&gt;XLS-65&lt;/a&gt;) is included in xrpld 3.4.0. It introduces three changes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;cash-basis accounting for loan brokers&lt;/li&gt;
&lt;li&gt;a closed-ended vault type&lt;/li&gt;
&lt;li&gt;stricter vault and loan invariants&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;All three affect the core lending transactions, &lt;code&gt;LoanSet&lt;/code&gt; and &lt;code&gt;LoanPay&lt;/code&gt;. We therefore assessed the amendment's performance impact as part of the 3.4.0 release.&lt;/p&gt;

&lt;h2&gt;
  
  
  Summary
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;LendingProtocolV1_1&lt;/code&gt; introduces no measurable performance regression.&lt;/strong&gt; Across about 20 A/B comparisons with the amendment enabled, we observed no regression in:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;throughput&lt;/li&gt;
&lt;li&gt;ledger close time&lt;/li&gt;
&lt;li&gt;per-transaction processing cost&lt;/li&gt;
&lt;li&gt;system resource utilization&lt;/li&gt;
&lt;/ul&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;3.3.0 → 3.4.0&lt;/th&gt;
&lt;th&gt;Lending (3 h)&lt;/th&gt;
&lt;th&gt;Mixed payments + lending (3 h)&lt;/th&gt;
&lt;th&gt;Endurance (12 h)*&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Transactions per cluster&lt;/td&gt;
&lt;td&gt;1.56 M&lt;/td&gt;
&lt;td&gt;1.56 M&lt;/td&gt;
&lt;td&gt;4.8 M&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;TPS&lt;/td&gt;
&lt;td&gt;149.31 → 149.28&lt;/td&gt;
&lt;td&gt;149.30 → 149.32&lt;/td&gt;
&lt;td&gt;111.99 → 111.99&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Ledger close mean (s)&lt;/td&gt;
&lt;td&gt;3.946 → 3.945&lt;/td&gt;
&lt;td&gt;4.126 → 4.120&lt;/td&gt;
&lt;td&gt;3.883 → 3.879&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Ledger close P95 (s)&lt;/td&gt;
&lt;td&gt;4.755 → 4.765&lt;/td&gt;
&lt;td&gt;5.050 → 5.070&lt;/td&gt;
&lt;td&gt;4.475 → 4.461&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Result&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;em&gt;*The endurance test runs a lighter version of the mixed payments + lending workload (about 15% &lt;code&gt;LoanPay&lt;/code&gt;) at a fixed 112 TPS to evaluate long-term stability.&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Methodology
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Test environment.&lt;/strong&gt; A private XRPL network sized to MainNet specifications:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;5 validator nodes&lt;/li&gt;
&lt;li&gt;4 client nodes that accept transactions from the load generators&lt;/li&gt;
&lt;li&gt;4 load generators&lt;/li&gt;
&lt;li&gt;Every test starts from the same ledger state: about 250,000 funded accounts with pre-built loan brokers, vaults and loans.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;A/B comparison.&lt;/strong&gt; Two identical clusters run side by side under the same load. The baseline cluster runs xrpld 3.3.0. The target cluster runs the new code (individual feature builds, then xrpld 3.4.0) with &lt;code&gt;LendingProtocolV1_1&lt;/code&gt; enabled, so the new vault and invariant checks apply to every lending transaction.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Evaluation.&lt;/strong&gt; For each test, we compared seven per-ledger metrics between the two clusters, including ledger close time, per-transaction processing time and consensus timing, using statistical tests to separate real changes from normal run-to-run variation.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Workloads.&lt;/strong&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Workload&lt;/th&gt;
&lt;th&gt;Transaction mix&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Lending&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;LoanSet&lt;/code&gt; (40%) and &lt;code&gt;LoanPay&lt;/code&gt; (60%) across IOU, MPT and XRP loans&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Payments&lt;/td&gt;
&lt;td&gt;XRP, IOU and cross-currency path payments&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Mixed payments + lending&lt;/td&gt;
&lt;td&gt;Direct and multi-path payments combined with &lt;code&gt;LoanPay&lt;/code&gt;, representing a production-like mix&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Results
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Feature-Level Testing
&lt;/h3&gt;

&lt;p&gt;We assessed each lending feature individually against 3.3.0, and then all features combined.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Feature&lt;/th&gt;
&lt;th&gt;Workload&lt;/th&gt;
&lt;th&gt;TPS&lt;/th&gt;
&lt;th&gt;Success rate&lt;/th&gt;
&lt;th&gt;Result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Cash-basis accounting&lt;/td&gt;
&lt;td&gt;Lending&lt;/td&gt;
&lt;td&gt;168&lt;/td&gt;
&lt;td&gt;99.999%&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Closed-ended vault&lt;/td&gt;
&lt;td&gt;Lending&lt;/td&gt;
&lt;td&gt;168&lt;/td&gt;
&lt;td&gt;99.998%&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Closed-ended vault&lt;/td&gt;
&lt;td&gt;Payments&lt;/td&gt;
&lt;td&gt;149&lt;/td&gt;
&lt;td&gt;99.996%&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;All lending features combined&lt;/td&gt;
&lt;td&gt;Lending&lt;/td&gt;
&lt;td&gt;168&lt;/td&gt;
&lt;td&gt;99.999%&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The payments test confirms that the updated invariant checks, which apply to all transaction types, do not affect non-lending transactions.&lt;/p&gt;

&lt;h3&gt;
  
  
  Lending Transactions on 3.4.0
&lt;/h3&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Transaction type&lt;/th&gt;
&lt;th&gt;Transactions&lt;/th&gt;
&lt;th&gt;Success rate&lt;/th&gt;
&lt;th&gt;Avg latency (ms), 3.3.0 → 3.4.0&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;
&lt;code&gt;LoanPay&lt;/code&gt; (IOU, MPT, XRP)&lt;/td&gt;
&lt;td&gt;936,708&lt;/td&gt;
&lt;td&gt;99.999%&lt;/td&gt;
&lt;td&gt;12.51 → 12.52&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;
&lt;code&gt;LoanSet&lt;/code&gt; (IOU, MPT, XRP)&lt;/td&gt;
&lt;td&gt;622,943&lt;/td&gt;
&lt;td&gt;99.999%&lt;/td&gt;
&lt;td&gt;25.76 → 23.01&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h3&gt;
  
  
  12-Hour Endurance
&lt;/h3&gt;

&lt;p&gt;Over 12 hours of sustained load, ledger close time stayed consistent with the 3.3.0 baseline (see summary table). Memory usage followed the same pattern on both versions: it rose during the first few hours as caches warmed up, then held steady. Throughout the test, the two versions stayed within 0.1 GB of each other.&lt;/p&gt;

&lt;h2&gt;
  
  
  Resource Utilization
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Resource&lt;/th&gt;
&lt;th&gt;Result&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;CPU (validators and client nodes)&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Memory&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Network&lt;/td&gt;
&lt;td&gt;No regression&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Threads, open files, disk I/O&lt;/td&gt;
&lt;td&gt;No change&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;CPU and memory differences between 3.3.0 and 3.4.0 averaged about 1%, comparable to the run-to-run variation we measured between identical clusters. All nodes operated well below capacity throughout testing.&lt;/p&gt;

&lt;h4&gt;
  
  
  CPU
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvy5uzb4rnicm8trkaghc.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fvy5uzb4rnicm8trkaghc.png" alt=" " width="800" height="141"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h4&gt;
  
  
  Memory
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzwmz6nz35z9727rh7vmx.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzwmz6nz35z9727rh7vmx.png" alt=" " width="798" height="144"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;The assessment covered feature-level testing, 3-hour comparisons on two workloads, and 12-hour endurance testing. &lt;code&gt;LendingProtocolV1_1&lt;/code&gt; showed no performance regression in any of these areas.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;From a performance standpoint, &lt;code&gt;LendingProtocolV1_1&lt;/code&gt; is ready for MainNet activation.&lt;/strong&gt;&lt;/p&gt;

</description>
      <category>xrpl</category>
      <category>blockchain</category>
      <category>web3</category>
    </item>
    <item>
      <title>Unlocking Institutional Finance on the XRP Ledger.</title>
      <dc:creator>Shota Natenadze</dc:creator>
      <pubDate>Thu, 08 Oct 2026 12:52:51 +0000</pubDate>
      <link>https://dev.to/ripplexdev/unlocking-collateral-mobility-on-xrpl-2j68</link>
      <guid>https://dev.to/ripplexdev/unlocking-collateral-mobility-on-xrpl-2j68</guid>
      <description>&lt;p&gt;&lt;em&gt;How Permission Delegation, Confidential Transfers, Atomic Batch, Sponsored Fees and Dynamic MPTs come together&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;It is Sunday evening and a tariff announcement drops. If you run an event-driven strategy, you already know what it means. Oil falls at Monday's open, and there is real opportunity in being early.&lt;/p&gt;

&lt;p&gt;You are holding $50 million in a tokenized assets. But it is Sunday. The clearing house is closed. Your bank is closed. Nobody is picking up. The tokens are sitting right there on your balance sheet and there is nothing you can do with it until someone in another time zone opens a laptop.&lt;/p&gt;

&lt;p&gt;By the time they do, the market has already moved. You were right, and it did not matter.&lt;/p&gt;

&lt;p&gt;Now run the same evening on the XRP Ledger. At 9pm on Sunday you pledge the tokens for stablecoins. Collateral and stablecoin change hands in the same instant, in about five seconds. No clearing house, no broker on call, no waiting for a payment rail to wake up.&lt;/p&gt;

&lt;p&gt;Nothing about the asset changed between those two versions of Sunday. It was already tokenized in both. What changed is that in the second one, you could actually use it.&lt;/p&gt;

&lt;p&gt;That is the difference between tokenizing an asset and mobilizing it, and it is where most of the industry is currently stuck. There are billions of dollars of tokenized funds, treasuries and money market instruments sitting on ledgers right now, doing roughly what they did on a spreadsheet. &lt;/p&gt;

&lt;p&gt;The reason is not the technology of tokenizing. It is everything that has to happen afterwards. A regulated institution cannot move an asset unless it can control who inside the firm is allowed to move it, keep the token accurate as the underlying fund changes daily, guarantee that cash and collateral move together or not at all, keep its trading strategy off a public feed, and do all of it without holding a crypto asset its own rules forbid it from touching.&lt;/p&gt;

&lt;p&gt;Public ledgers have not been able to offer all five at once. That is the actual blocker, and it has very little to do with blockchains being fast or cheap. Five features released under &lt;a href="https://xrpl.org/blog/2026/xrpld-3.3.0" rel="noopener noreferrer"&gt;rippled 3.3.0&lt;/a&gt; close those gaps one by one. The clearest way to see how, is to follow a single asset through its life, so the rest of this piece follows a hypothetical bank, &lt;strong&gt;XBank&lt;/strong&gt;, as it issues a tokenized money market fund and takes it all the way to that Sunday night trade.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Issue and Delegate&lt;/strong&gt;&lt;br&gt;
XBank tokenizes its money market fund as XMMF using XRPL's Multi-Purpose Token standard. But XBank's internal governance requires them to distribute responsibilities based on internal risk matrix.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://opensource.ripple.com/docs/xls-75-permission-delegation" rel="noopener noreferrer"&gt;Permission Delegation&lt;/a&gt; is XRPL's native answer to role-based access: where one account can grant another account a scoped permission to act on its behalf, all enforced by the ledger, instead of an application layer. In XBank's case, treasury delegates a narrow capability to the trading desk: the ability to send XMMF payments. The desk can transact within that scope but cannot mint, burn, or modify governance.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fnph8p47w3sg121eu8nd4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fnph8p47w3sg121eu8nd4.png" alt=" " width="800" height="450"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Keep the Assets Current&lt;/strong&gt;&lt;br&gt;
XBank has gone through a change in regulation that impacts their prospectus, which requires them to revise their prospectus URI.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://opensource.ripple.com/docs/xls-94-dynamic-mpts/dynamic-mpts" rel="noopener noreferrer"&gt;Dynamic MPTs&lt;/a&gt; extend the MPT standard by letting issuers mark specific token properties as updatable at the moment of creation, while everything else stays permanently locked. XBank uses this to declare that in the metadata and select compliance flags can evolve post-issuance. The mutability boundary is transparent on-ledger, and it is one-way: once a flag is enabled, it cannot be disabled. Counterparties know exactly what can change and what cannot. &lt;/p&gt;

&lt;p&gt;This matters for the flow that follows. When XBank's trading desk swaps tokenised assets for stabelcoins, the counterparty needs confidence that the token they are receiving reflects the fund's current state, not an outdated snapshot. Dynamic MPTs keep the on-chain representation aligned with the off-chain reality of the fund, without reissuing.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Make it Private&lt;/strong&gt;&lt;br&gt;
&lt;a href="https://opensource.ripple.com/docs/xls-96-confidential-transfers" rel="noopener noreferrer"&gt;Confidential Transfers&lt;/a&gt; add a privacy layer to MPT transfers: amounts are encrypted on-ledger using zero-knowledge proofs, so the ledger can verify that balances are sufficient without ever seeing the actual figures. Observers see that a transfer occurred but cannot read the amount.&lt;/p&gt;

&lt;p&gt;What makes the design workable for institutions is selective decryption. The sender holds a decryption key that can be shared with specific parties: internal auditors, external auditors, the regulator. Public observers see nothing; permissioned reviewers see exactly what they need for quarterly reporting, tax filings, or supervisory review. Confidentiality is also opt-in per transaction, so a fund can keep the collateral leg private while leaving a regulated cash leg transparent, if its compliance posture calls for it.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3ixgipbn5sjkfvhwd0n6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3ixgipbn5sjkfvhwd0n6.png" alt=" " width="800" height="450"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4.Abstract the Fees&lt;/strong&gt;&lt;br&gt;
Every XRPL transaction requires a small XRP fee. For a regulated fund, holding XRP, directly or indirectly, is not permitted under Irish, UK, or most EU law. The same blocker hits US '40 Act funds, pension funds, and insurers. The technical requirement to transact contradicts the legal requirement to operate.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://opensource.ripple.com/docs/xls-68-sponsored-fees-and-reserves" rel="noopener noreferrer"&gt;Sponsored Fees and Reserves&lt;/a&gt; allow one XRPL account to cover the transaction fees and reserve requirements on behalf of another, removing the need for the transacting account to hold any XRP at all. For XBank, a third-party sponsor (the marketplace, a tokenization provider) pays the XRP fees on behalf of the fund. The fund never holds or touches XRP. From its accounting and compliance perspective, it operates exclusively in regulated assets. This is the same model that lets a TradFi fund use SWIFT without paying SWIFT directly: a service layer underneath that the fund never sees. &lt;/p&gt;

&lt;p&gt;The Aviva pilot, currently in flight, is the first production-scale test of this pattern and every institutional MMF and RWA program coming to XRPL will be a natural beneficiary of fees sponsorship at the first transaction.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhqhc1tb71b3abkvgwdby.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhqhc1tb71b3abkvgwdby.png" alt=" " width="800" height="450"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Looking forward&lt;/strong&gt;&lt;br&gt;
Together, all five features they move XRPL past issuance into the phase that actually matters: collateral mobility. A regulated institution can issue a tokenized asset, keep it current, trade it atomically and privately, and do all of it without holding a crypto asset or breaking its compliance posture. Multiple financial institutions are lined up behind this release like Aviva and Licuido. &lt;/p&gt;

</description>
      <category>xrpl</category>
      <category>atomicity</category>
      <category>collateral</category>
      <category>tokenization</category>
    </item>
    <item>
      <title>LendingProtocolV1_1: Closed-Ended Vaults and Cash-Basis Accounting</title>
      <dc:creator>Shota Natenadze</dc:creator>
      <pubDate>Wed, 30 Sep 2026 15:39:48 +0000</pubDate>
      <link>https://dev.to/ripplexdev/lendingprotocolv11-closed-ended-vaults-and-cash-basis-accounting-2me8</link>
      <guid>https://dev.to/ripplexdev/lendingprotocolv11-closed-ended-vaults-and-cash-basis-accounting-2me8</guid>
      <description>&lt;p&gt;The Single Asset Vault has two modes: open-ended and closed-ended. Both use the same Vault ledger object and the same transaction types (&lt;code&gt;VaultCreate, VaultDeposit, VaultWithdraw, VaultSet&lt;/code&gt;). The difference is when capital can move in and out.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Open-ended vaults&lt;/strong&gt; accept deposits and process withdrawals at any time. When a depositor sends assets, they receive shares (&lt;code&gt;MPTs&lt;/code&gt;) at the current price per share (&lt;code&gt;PPS = AssetsTotal / SharesTotal&lt;/code&gt;). When they withdraw, shares are burned and assets returned at whatever PPS is at that moment. Capital flows are continuous. Open-ended vaults are fully usable for non-lending purposes (e.g., pooled custody, shared treasury), but loan brokers cannot be attached to them.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Closed-ended vaults&lt;/strong&gt; (up for validator voting under the &lt;a href="https://livenet.xrpl.org/amendment/A360E2BFD775A5B0DCE1C36C16DF31B72735A57584FD163655D2F9564F8E7AC8" rel="noopener noreferrer"&gt;LendingProtocolV1_1&lt;/a&gt; amendment) have a fixed lifecycle with three phases:&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. Subscription.&lt;/strong&gt; The vault is open for deposits. Shares are minted. The vault owner sets the subscription window via &lt;code&gt;SubscriptionDate&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Investment.&lt;/strong&gt; The vault closes to new deposits (the protocol enforces this automatically). The vault owner deploys capital into loans. Interest is returned to the vault via &lt;code&gt;VaultDeposit&lt;/code&gt; with the &lt;code&gt;tfVaultDonation&lt;/code&gt; flag, which increases &lt;code&gt;AssetsTotal&lt;/code&gt; without minting new shares, raising PPS for all existing holders.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Redemption.&lt;/strong&gt; After &lt;code&gt;RedemptionDate&lt;/code&gt;, the vault reopens for withdrawals. Depositors redeem shares at the final PPS (principal plus earned yield).&lt;/p&gt;

&lt;p&gt;Closed-ended vaults are created by setting &lt;code&gt;VaultKind = 1&lt;/code&gt; on &lt;code&gt;VaultCreate&lt;/code&gt;along with &lt;code&gt;SubscriptionDate&lt;/code&gt; and &lt;code&gt;RedemptionDate&lt;/code&gt;. This field is immutable after creation.&lt;/p&gt;

&lt;p&gt;Under LendingProtocolV1_1, LoanBrokerSet requires the vault to be closed-ended and returns &lt;code&gt;tecNO_PERMISSION&lt;/code&gt; on an open-ended vault. Nothing else about open-ended vaults changes: they still exist and still accept deposits and withdrawals at any time. The only thing blocked is setting up a loan broker on them.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What LendingProtocolV1_1 introduces&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;The &lt;a href="https://xrpl.org/resources/known-amendments#lendingprotocolv1_1" rel="noopener noreferrer"&gt;LendingProtocolV1_1 amendment&lt;/a&gt; extends the &lt;code&gt;LendingProtocol&lt;/code&gt; and &lt;code&gt;SingleAssetVault&lt;/code&gt; amendments with two changes:&lt;/p&gt;

&lt;p&gt;First, it introduces the closed-ended vault as a new vault type that users can create. Users can still create both open-ended and closed-ended vaults, but after this amendment is enabled, new loan brokers can only be created against closed-ended vaults. &lt;/p&gt;

&lt;p&gt;Second, it introduces cash-basis accounting. Under the original LendingProtocol amendment, all scheduled interest on a loan was recognized at the time of loan origination: the moment the loan was created, the full expected interest was recorded as income in the vault's accounting. &lt;code&gt;LendingProtocolV1_1&lt;/code&gt; changes this so that interest is only recognized as income when payments are actually made. This is a more conservative and accurate accounting model, and it means the vault's &lt;code&gt;AssetsTotal&lt;/code&gt; only reflects interest that has actually been received, not interest that is expected in the future.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Why the Closed and Open-Ended Vaults ship together&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Lending is restricted to closed-ended vaults because of how share pricing works. In an open-ended vault, anyone can enter or exit at any time, so a depositor could time their entry and exit around a change in price per share and capture yield they did not earn. Closed-ended vaults avoid this by design: once the subscription window closes, no new shares can be minted, so there is no way to enter mid-term.&lt;/p&gt;

&lt;p&gt;Enabling lending on open-ended vaults would require a mechanism that prevents this kind of timing capture under continuous entry and exit. That is open design work, to be scheduled for an upcoming release.&lt;/p&gt;

&lt;p&gt;The decision was to ship the closed-ended vault first and require it to activate together with the original Lending Protocol and the Single Asset Vault.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What this means for the amendment vote&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Three amendments need to activate for lending to go live on mainnet:&lt;br&gt;
&lt;a href="https://livenet.xrpl.org/amendment/A360E2BFD775A5B0DCE1C36C16DF31B72735A57584FD163655D2F9564F8E7AC8" rel="noopener noreferrer"&gt;LendingProtocolV1_1&lt;/a&gt; goes together with the &lt;a href="https://livenet.xrpl.org/amendment/565B90CA1AB2B9D42208ED10884188C64F9E19083DECB9634AAF06EB03299509" rel="noopener noreferrer"&gt;LendingProtocol&lt;/a&gt; and &lt;a href="https://livenet.xrpl.org/amendment/81BD2619B6B3C8625AC5D0BC01DE17F06C3F0AB95C7C87C93715B87A4FD240D8" rel="noopener noreferrer"&gt;SingleAssetVault&lt;/a&gt;. &lt;code&gt;LendingProtocol&lt;/code&gt; is the base lending engine: loan brokers, loan origination, repayment, default handling, and first-loss capital. &lt;code&gt;SingleAssetVault&lt;/code&gt; is the base vault primitive that both open-ended and closed-ended vaults build on. &lt;/p&gt;

</description>
      <category>architecture</category>
      <category>blockchain</category>
      <category>xls66</category>
      <category>xls65</category>
    </item>
    <item>
      <title>Confidential Transfer - QA Test Report</title>
      <dc:creator>Ramkumar SG</dc:creator>
      <pubDate>Tue, 22 Sep 2026 21:19:23 +0000</pubDate>
      <link>https://dev.to/ripplexdev/confidential-transfer-qa-test-report-1l18</link>
      <guid>https://dev.to/ripplexdev/confidential-transfer-qa-test-report-1l18</guid>
      <description>&lt;p&gt;&lt;strong&gt;Test Report Date&lt;/strong&gt;: 9/22/2026&lt;br&gt;
&lt;strong&gt;Prepared By&lt;/strong&gt;: &lt;a href="https://github.com/sgramkumar" rel="noopener noreferrer"&gt;sgramkumar&lt;/a&gt;&lt;br&gt;
&lt;strong&gt;Environment&lt;/strong&gt;: GitLab CI Runner (nix-debian)&lt;br&gt;
&lt;strong&gt;Network&lt;/strong&gt;: xrpld devnet + private CI network&lt;/p&gt;
&lt;h2&gt;
  
  
  Overview
&lt;/h2&gt;

&lt;p&gt;This report presents the results of QA testing performed on Confidential Transfer (XLS-0096) across xrpld servers. Coverage targets the Confidential MPT feature, which layers ElGamal-encrypted balances and zero-knowledge proofs on top of Multi-Purpose Tokens so that token amounts can be held and transferred confidentially while remaining verifiable on-ledger.&lt;/p&gt;
&lt;h2&gt;
  
  
  1. Feature
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Feature Name&lt;/strong&gt;: Confidential Transfer (Confidential MPT)&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Description&lt;/strong&gt;: Standard Multi-Purpose Tokens expose every balance and transfer amount in clear on the ledger, which is unacceptable for many payment and settlement use cases. Confidential MPT introduces a confidential balance model in which amounts are stored as ElGamal ciphertexts and every state transition is accompanied by a zero-knowledge proof that the transition is valid without revealing the underlying value. The feature adds five transactors — &lt;code&gt;ConfidentialMPTConvert&lt;/code&gt; (move a public MPT balance into the confidential spending balance), &lt;code&gt;ConfidentialMPTSend&lt;/code&gt; (transfer a confidential amount into a destination's inbox), &lt;code&gt;ConfidentialMPTMergeInbox&lt;/code&gt; (fold received inbox ciphertexts into the spending balance), &lt;code&gt;ConfidentialMPTConvertBack&lt;/code&gt; (move a confidential balance back to public), and &lt;code&gt;ConfidentialMPTClawback&lt;/code&gt; (issuer clawback of confidential balances). An optional Auditor may be configured on the issuance so a designated party can decrypt amounts for compliance, and the confidential surface composes with the existing issuance-control flags (lock, freeze, require-auth, clawback).&lt;/p&gt;

&lt;p&gt;Specification Reference: &lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0096-confidential-mpt" rel="noopener noreferrer"&gt;https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0096-confidential-mpt&lt;/a&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  2. Test Scope
&lt;/h2&gt;

&lt;p&gt;This report covers the full Confidential Transfer surface as it stands under the Confidential MPT amendment. Testing focused on ensuring that:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;All five transactors (Convert, Send, MergeInbox, ConvertBack, Clawback) behave per the XLS-0096 specification, including correct balance conservation and commit/rollback semantics&lt;/li&gt;
&lt;li&gt;ElGamal ciphertext handling, zero-knowledge proof verification, per-send re-randomization, and the caller-specified discrete-log decrypt range behave correctly and reject malformed or stale material&lt;/li&gt;
&lt;li&gt;Issuance lifecycle and control flags (privacy flag, can-lock/lock, global lock, require-auth, clawback, auditor configuration) are enforced across every transactor&lt;/li&gt;
&lt;li&gt;The auditor mirror is maintained and re-randomized correctly so that a configured auditor can always decrypt while confidentiality is preserved against everyone else&lt;/li&gt;
&lt;li&gt;Balance and supply invariants hold — encrypted balance deltas match public balance deltas, inbox and spending balances reconcile, and no value is created or destroyed&lt;/li&gt;
&lt;li&gt;Security-critical properties hold under adversarial input, including bad proofs, mismatched ciphertexts, stale-proof replay, crafted randomness, and structural/serialization attacks&lt;/li&gt;
&lt;li&gt;Confidential MPT composes correctly with Batch, Permission Delegation, Sponsored Fees &amp;amp; Reserves, Deposit Authorization / Preauth, and Permissioned Domains&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;
  
  
  3. Types of Testing Conducted
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Testing Type&lt;/th&gt;
&lt;th&gt;Description&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Functional&lt;/td&gt;
&lt;td&gt;Verifying each transactor, issuance lifecycle step, control flag, and RPC/metadata surface against the XLS-0096 specification&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cryptographic&lt;/td&gt;
&lt;td&gt;ElGamal ciphertext validation, zero-knowledge proof verification, per-send re-randomization, auditor-mirror correctness, and the discrete-log decrypt-range API&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Regression&lt;/td&gt;
&lt;td&gt;Running the full xrpld test suite to confirm Confidential MPT changes did not break existing functionality&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Adversarial / Security&lt;/td&gt;
&lt;td&gt;Attack-matrix scenarios probing bad proofs, mismatched/stale ciphertexts, crafted randomness, key rotation, and structural/serialization smuggling&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cross-Feature&lt;/td&gt;
&lt;td&gt;Interactions between Confidential MPT and Batch (XLS-56), Permission Delegation (XLS-75), Sponsored Fees &amp;amp; Reserves, Deposit Authorization / Preauth, and Permissioned Domains&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;End-to-End&lt;/td&gt;
&lt;td&gt;Full flows spanning convert, confidential send, inbox merge, convert-back, and clawback, with downstream balance and metadata validation&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;
&lt;h2&gt;
  
  
  4. Test Results Summary
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Testing Type&lt;/th&gt;
&lt;th&gt;Total Tests&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Confidential Transfer — Core Functional (five transactors, issuance lifecycle, crypto, invariants)&lt;/td&gt;
&lt;td&gt;358&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Confidential Transfer — Adversarial / Security (adversarial, hardening, bug hunt)&lt;/td&gt;
&lt;td&gt;110&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Confidential Transfer — Cross-Feature (Batch/Composition, Delegation, Sponsor, Deposit Preauth, Permissioned Domain)&lt;/td&gt;
&lt;td&gt;93&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Confidential Transfer — Total&lt;/td&gt;
&lt;td&gt;561&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Regression — xrpld (full suite)&lt;/td&gt;
&lt;td&gt;5,088&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Testcases: &lt;a href="https://dev.to/ripplexdev/confidential-transfer-testcases-1pee"&gt;https://dev.to/ripplexdev/confidential-transfer-testcases-1pee&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Feature commit history&lt;/strong&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight markdown"&gt;&lt;code&gt;&lt;span class="p"&gt;-&lt;/span&gt; Introduced as the ConfidentialTransfer amendment in PR #5860 (768d7603b1, merged 2026-06-27) 
&lt;span class="p"&gt;-&lt;/span&gt; Specification standardized as XLS-0096 (XRPL-Standards discussion #372, spec created 2026-01-15) and tracked through spec rc2 (#303)
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Related xrpld changes covered in this round:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/7279" rel="noopener noreferrer"&gt;#7279&lt;/a&gt; — Add Batch + Delegation + Confidential tests&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/7535" rel="noopener noreferrer"&gt;#7535&lt;/a&gt; — per-send re-randomization of zero-knowledge proof scalars (independent re-randomization scalar per send)&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/7612" rel="noopener noreferrer"&gt;#7612&lt;/a&gt; — auditor-mirror re-randomization and the MergeInbox defense against crafted-randomness attacks&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/6988" rel="noopener noreferrer"&gt;#6988&lt;/a&gt; — DestinationTag handling on confidential sends&lt;/li&gt;
&lt;li&gt;mpt-crypto PR &lt;a href="https://github.com/XRPLF/mpt-crypto/pull/127" rel="noopener noreferrer"&gt;#127&lt;/a&gt; — caller-specified discrete-log decrypt range and the shared-secret zero-exclusion guard&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  5. Bugs Reported
&lt;/h2&gt;

&lt;p&gt;All internal bugs are fixed and there are no Critical Open bugs.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. Conclusion
&lt;/h2&gt;

&lt;p&gt;Confidential Transfer (XLS-0096) has now been exercised across 561 dedicated tests spanning core functional coverage (the five transactors, issuance lifecycle, cryptographic verification, and ledger invariants), adversarial / security scenarios, and cross-feature interactions with Batch, Permission Delegation, Sponsored Fees &amp;amp; Reserves, Deposit Authorization / Preauth, and Permissioned Domains. In addition, the full xrpld regression suite has been executed to confirm no downstream breakage.&lt;/p&gt;

&lt;p&gt;All security-critical properties documented in XLS-0096 have been validated, including:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Correct balance conservation across every transactor — encrypted balance deltas match the corresponding public balance deltas, and inbox/spending balances reconcile with no value created or destroyed&lt;/li&gt;
&lt;li&gt;Rejection of malformed, mismatched, and stale ciphertexts and zero-knowledge proofs, and of crafted-randomness attacks against the auditor mirror and MergeInbox (PR &lt;a href="https://github.com/XRPLF/rippled/pull/7612" rel="noopener noreferrer"&gt;#7612&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;Correct per-send re-randomization of proof scalars so that confidentiality holds across repeated and chained sends (PR &lt;a href="https://github.com/XRPLF/rippled/pull/7535" rel="noopener noreferrer"&gt;#7535&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;Correct enforcement of issuance control flags (privacy, lock/global-lock, require-auth, clawback) and auditor configuration across all five transactors&lt;/li&gt;
&lt;li&gt;Correct interaction with delegated authority, sponsorship, tickets, deposit authorization, permissioned domains, and Batch composition, including atomic rollback of stale-proof inner transactions&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The feature is considered ready for production use at the tested commit level. Future updates to this report will follow any material additions to the Confidential Transfer surface.&lt;/p&gt;

</description>
      <category>blockchain</category>
      <category>crypto</category>
      <category>security</category>
      <category>testing</category>
    </item>
    <item>
      <title>Confidential Transfer - Testcases</title>
      <dc:creator>Ramkumar SG</dc:creator>
      <pubDate>Tue, 22 Sep 2026 21:17:39 +0000</pubDate>
      <link>https://dev.to/ripplexdev/confidential-transfer-testcases-1pee</link>
      <guid>https://dev.to/ripplexdev/confidential-transfer-testcases-1pee</guid>
      <description>&lt;p&gt;Automated test coverage for the Confidential Transfer feature (&lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0096-confidential-mpt" rel="noopener noreferrer"&gt;XLS-0096&lt;/a&gt;), grouped by the transactor, invariant, or cross-feature interaction each test exercises with 561 tests.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Category&lt;/th&gt;
&lt;th&gt;Test Count&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Transactor — Convert&lt;/td&gt;
&lt;td&gt;72&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Transactor — Send&lt;/td&gt;
&lt;td&gt;69&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Transactor — MergeInbox&lt;/td&gt;
&lt;td&gt;31&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Transactor — ConvertBack&lt;/td&gt;
&lt;td&gt;30&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Transactor — Clawback&lt;/td&gt;
&lt;td&gt;30&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Issuance Lifecycle&lt;/td&gt;
&lt;td&gt;41&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Crypto&lt;/td&gt;
&lt;td&gt;44&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hardening&lt;/td&gt;
&lt;td&gt;36&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Adversarial / Security&lt;/td&gt;
&lt;td&gt;45&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Invariants&lt;/td&gt;
&lt;td&gt;41&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Batch / Composition&lt;/td&gt;
&lt;td&gt;18&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Delegation&lt;/td&gt;
&lt;td&gt;12&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cross-Feature&lt;/td&gt;
&lt;td&gt;63&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Bug Hunt&lt;/td&gt;
&lt;td&gt;29&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Total&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;561&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Transactor — Convert
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Confidential MPT convert basic&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert without tfMPTCanPrivacy flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert partial amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert multiple times with available balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert multiple times without ignoring holder elgamal public key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert multiple times with insufficient balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert account is individual account locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert account is globally locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor enabled and auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor same as issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor enabled separately after issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with different issuer elgamal public key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert holder and issuer encrypted amounts different&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor enabled but no auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor not enabled but with auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor enabled and using issuer for auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor enabled and using holder for auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with no issuer elgamal public key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert set privacy flag and issuer elgamal public key separately after MPT Issuance create&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert set privacy flag and issuer elgamal public key in same transaction after MPT Issuance create&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert existing account set privacy flag when oa greater than zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert new account after set privacy flag when oa greater than zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert set privacy flag when oa equals zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert set privacy flag by holder when oa greater than zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert set privacy flag by holder when oa equals zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with cannot mutate privacy flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with cannot mutate privacy flag and set privacy flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with cannot mutate privacy flag and set privacy flag again&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert holder without elgamal public key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert non existent account&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert account with no MPT balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert account with no MPT object&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with MPT lock flag set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with can lock flag but not locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert without tfMPTCanTransfer flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with clawback flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert from issuer account&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert from account who is another issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert zero amount account as issuer with balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert zero amount account as issuer with no balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert account as third party without MPT object&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert MPT amount more than holder balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert MPT amount in fraction&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert MPT amount zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert MPT amount negative&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert negative MPT amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert no holder encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert no issuer encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert no auditor encrypted amount when auditor set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with auditor encrypted amount but auditor not set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with receiver field&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with destination field&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with invalid MPT Issuance id&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert MPT amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert maximum amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert auditor key mismatch&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert invalid ciphertext format&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert invalid zkp&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert multiple accounts&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert change auditor&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert and delete holder&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with multisign with disabling master key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert with multisign without disabling master key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert minimum amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert invalid elgamal key format&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert elgamal key wrong length&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert auditor encrypted amount encrypts wrong value&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert issuer key rotation impact when coa is non zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert issuer key rotation impact when coa is zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert identity point elgamal key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert blinding factor wrong length&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert zero amount on existing confidential balance&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Transactor — Send
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Confidential MPT send to destination account basic&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send partial amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send multiple times&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send source destination swap&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with one drop in inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without source account merging inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without bob having mptoken&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with bob having zero mpt&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with bob having nonzero MPT but no coa&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to destination account having zero confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send zero amount to bob&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send source confidential balance is zero after convert&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to self&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send source does not exist&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send source does not have mptoken&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send source is locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send destination is nonexistent account&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send destination is locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with globally locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send more than available coa&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send invalid MPT Issuance id&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send issuer sends to holder&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send holder sends to issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send sender encrypted amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send sender encrypted amount invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send destination encrypted amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send destination encrypted amount invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send issuer encrypted amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with auditor basic success&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with auditor missing auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with auditor invalid auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without auditor but with auditor encrypted amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without auditor unexpected auditor amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with auditor verify decryption&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with auditor multiple sends&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send from multiple accounts&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with zkproof missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with invalid zkproof value&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with amount commitment missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with invalid amount commitment value&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with balance commitment missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with invalid balance commitment value&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send multiple times in sequence&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send version increment&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send incorrect version&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send stale proof rejection&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send mismatched ciphertexts&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send and MPT Issuance destroy source account having zero confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send and delete mptoken on destination account before merging inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send second account as sender&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to different accounts with same version&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send first merge after convert merge convert&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with multisign without disabling master key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with multisign with disabling master key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with multisign insufficient signatures&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with regular key master enabled&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with regular key disabled master&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send fractional amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send negative amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send and public payment to same destination&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send and public payment to different destinations&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without merge&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with require auth and destination unauthorized&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to destination with deposit auth&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to destination with deposit auth and preauth&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send to destination requiring credential without providing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with nonexistent credential&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send with expired credential&lt;/li&gt;
&lt;li&gt;Test Confidential MPT send without tfMPTCanTransfer flag&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Transactor — MergeInbox
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Confidential MPT merge inbox basic&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox after partial convert&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox account does not exist&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox account does not have mptoken&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with no confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with various token balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox account is issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox convert twice then merge&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox account is individual account locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox account is globally locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox invalid MPT Issuance id&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox missing MPT Issuance id&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox by third party&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox empty&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox conver and merge multiple times&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox enczero determinism&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox enczero validation&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox and destroy mptoken without merge inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox and destroy mptoken after merge inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with maximum value&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox rapid state changes&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with mixed public confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox at minimum xrp reserve&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with require auth flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox idempotency empty inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox determinism across accounts&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox balance conservation&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox decrypt verification&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox with multisign with disabled master key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox version increments by one&lt;/li&gt;
&lt;li&gt;Test Confidential MPT merge inbox cb in amount field mismatch adversarial&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Transactor — ConvertBack
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Confidential MPT convert back entire confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back partial confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back without merging inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back with balance in inbox and spending&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back with balance only in inbox&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back with no balance in both inbox and spending&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back account does not exist&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back account does not have mptoken&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back account is issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back invalid mptokenissuanceid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back mptamount greater than spending balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back mptamount invalid values&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back zero amount&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back mptamount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back holder encrypted amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back holder encrypted amount invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back issuer encrypted amount missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back issuer encrypted amount invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back blinding factor missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back blinding factor invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back zkproof missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back zkproof invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back balance commitment missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back balance commitment invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back with auditor encrypted amount no auditor set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back no auditor encrypted amount with auditor set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back with auditor encrypted amount and auditor set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back account locked&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back and destroy mptoken and delete accounts&lt;/li&gt;
&lt;li&gt;Test Confidential MPT convert back encrypted balance delta equals public balance delta&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Transactor — Clawback
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Confidential MPT clawback with tfMPTCanClawback set&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback after partial convert&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback public mpt&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback without tfMPTCanClawback flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback non issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback holder missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback holder nonexistent&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback third party holder&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback MPT amount zero&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback from holder with zero confidential balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback MPT amount exceeds balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback MPT amount less than balance&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback zkproof missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback zkproof invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback MPT Issuance id missing&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback MPT Issuance id invalid&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback without privacy flag&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback without issuer encryption key&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback multiple holders&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback holder equals issuer&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback with inbox and spending&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback from source after send&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback from destination after send&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback public balance with regular clawback while confidential setup exists&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback dual public and confidential workflow&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback and delete account&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback with auditor&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback holder account individually locked before clawback&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback tokens globally locked before clawback&lt;/li&gt;
&lt;li&gt;Test Confidential MPT clawback invariant coa decreases&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Issuance Lifecycle
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Issuance set set confidential when coa gt zero is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set clear confidential bit is unrecognized returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Issuance set set confidential when cannot mutate is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set replace issuer key is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set replace auditor key surfaces as temMALFORMED&lt;/li&gt;
&lt;li&gt;Test Issuance set keys when confidential not enabled is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set keys with enable in same tx succeeds&lt;/li&gt;
&lt;li&gt;Test Issuance set keys with unrecognized clear bit is temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Issuance set re register issuer key after coa gt zero is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set set confidential via mutable flag sets lsf flag&lt;/li&gt;
&lt;li&gt;Test Issuance set confidential mutate with holder field is temMALFORMED&lt;/li&gt;
&lt;li&gt;Test Issuance set keys with holder field is temMALFORMED&lt;/li&gt;
&lt;li&gt;Test Issuance set confidential with nonzero transfer fee is tecNO PERMISSION&lt;/li&gt;
&lt;li&gt;Test Issuance set enable privacy and nonzero transfer fee same tx temBAD TRANSFER FEE&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create privacy flag without issuer key&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create issuer key without privacy flag&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create auditor key without privacy flag&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create invalid issuer elgamal key&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create invalid auditor elgamal key&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create coa initialized to zero&lt;/li&gt;
&lt;li&gt;Test MPT Issuance set remove auditor key with active supply&lt;/li&gt;
&lt;li&gt;Test MPT Issuance set non issuer privacy settings&lt;/li&gt;
&lt;li&gt;Test MPT Issuance create privacy flag with nonzero transfer fee&lt;/li&gt;
&lt;li&gt;Test MPT Issuance set auditor key without issuer key&lt;/li&gt;
&lt;li&gt;Test MPT Issuance destroy with nonzero confidential outstanding amount&lt;/li&gt;
&lt;li&gt;Test Privacy enabled at creation cannot be disabled by any clear bit&lt;/li&gt;
&lt;li&gt;Test Privacy enabled via mutable flag post creation then clear attempt returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Privacy frozen off via tfMPTCannotMutatePrivacy blocks all later enables&lt;/li&gt;
&lt;li&gt;Test Re enable privacy on already enabled Issuance is idempotent&lt;/li&gt;
&lt;li&gt;Test All six capability flags one way under confidential issuance&lt;/li&gt;
&lt;li&gt;Test Enabling privacy post creation via mutableflags enables all confidential transactors&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle failed unauthorize does not decrement coa&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle unauthorize after full round trip succeeds&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle unauthorize when other holder has cb succeeds&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle reauthorize after unauthorize yields fresh mptoken&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle issuer unauthorize blocks subsequent sends&lt;/li&gt;
&lt;li&gt;Test Authorize lifecycle authorize destroyed Issuance returns object not found&lt;/li&gt;
&lt;li&gt;Test Issuer can create confidential MPT with canclawback without account flag&lt;/li&gt;
&lt;li&gt;Test Issuer can confidential clawback without account flag&lt;/li&gt;
&lt;li&gt;Test Setting asfAllowTrustLineClawback after MPT Issuance returns tecOWNERS&lt;/li&gt;
&lt;li&gt;Test Issuance without tfMPTCanClawback rejects confidential clawback even with account flag&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Crypto
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Crypto wrapper generate keypair returns correctly sized hex&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper generate keypair is non deterministic&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper generate blinding factor is correctly sized and nondeterministic&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper encrypt decrypt round trip&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper decrypt with wrong privkey does not silently succeed&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper decrypt amount outside range returns rc minus 1&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper decrypt at exact range bounds inclusive&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper elgamal round trip is homomorphic consistent&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper same plaintext distinct blinding yields distinct ciphertexts&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper encrypt with explicit blinding factor is deterministic&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper pedersen commitment is deterministic in inputs&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper pedersen commitment differs on different amount&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper encrypt with wrong sized pubkey raises valueerror&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper decrypt with wrong sized privkey raises valueerror&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper encrypt with wrong sized blinding factor raises valueerror&lt;/li&gt;
&lt;li&gt;Test Crypto wrapper pedersen commitment with wrong sized bf raises valueerror&lt;/li&gt;
&lt;li&gt;Test Default range decrypts small amount&lt;/li&gt;
&lt;li&gt;Test Narrow range decrypts value within range&lt;/li&gt;
&lt;li&gt;Test Below range returns failure&lt;/li&gt;
&lt;li&gt;Test Above range returns failure&lt;/li&gt;
&lt;li&gt;Test Boundary low and high both decrypt&lt;/li&gt;
&lt;li&gt;Test Inverted range returns failure&lt;/li&gt;
&lt;li&gt;Test Zero excluded when range low gt zero&lt;/li&gt;
&lt;li&gt;Test Auditor decrypts real ciphertext with explicit range end to end&lt;/li&gt;
&lt;li&gt;Test Auditor decrypts initial convert balance to actual amount&lt;/li&gt;
&lt;li&gt;Test Auditor view decreases correctly after sender sends&lt;/li&gt;
&lt;li&gt;Test Auditor view increases correctly after receiver merges&lt;/li&gt;
&lt;li&gt;Test Auditor aggregate view equals Issuance confidential outstanding&lt;/li&gt;
&lt;li&gt;Test Auditor view unchanged across repeated rerandomization sends&lt;/li&gt;
&lt;li&gt;Test Auditor view zeros after full convert back&lt;/li&gt;
&lt;li&gt;Test Rerand destination inbox differs from tx ciphertext&lt;/li&gt;
&lt;li&gt;Test Rerand destination issuer mirror differs from tx ciphertext&lt;/li&gt;
&lt;li&gt;Test Rerand destination auditor mirror behavior&lt;/li&gt;
&lt;li&gt;Test Rerand sender spending balance is not rerandomized&lt;/li&gt;
&lt;li&gt;Test Rerand sender issuer mirror is not rerandomized&lt;/li&gt;
&lt;li&gt;Test Rerand multiple sends end to end decryption&lt;/li&gt;
&lt;li&gt;Test Rerand two consecutive sends produce different inbox deltas&lt;/li&gt;
&lt;li&gt;Test Rerand destination inbox changes under zero value send&lt;/li&gt;
&lt;li&gt;Test Rerand version counter increments once per send&lt;/li&gt;
&lt;li&gt;Test Rerand destination inbox and issuer mirror stay in lockstep per send&lt;/li&gt;
&lt;li&gt;Test Rerand send to self is rejected at preflight and does not mutate state&lt;/li&gt;
&lt;li&gt;Test Rerand two different senders to same destination compose correctly&lt;/li&gt;
&lt;li&gt;Test Rerand does not break issuer clawback path after send&lt;/li&gt;
&lt;li&gt;Test Rerand auditor mirror differs between two identical sends&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Hardening
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Send to destination requiring dest tag fails&lt;/li&gt;
&lt;li&gt;Test Send to destination requiring dest tag with tag succeeds&lt;/li&gt;
&lt;li&gt;Test Send with dest tag and auditor succeeds&lt;/li&gt;
&lt;li&gt;Test Send when destination authorized but never converted fails&lt;/li&gt;
&lt;li&gt;Test Send when sender authorized but never converted fails&lt;/li&gt;
&lt;li&gt;Test Send receiver confidential version unchanged&lt;/li&gt;
&lt;li&gt;Test Convert zero amount from individually frozen account fails&lt;/li&gt;
&lt;li&gt;Test Convert zero amount from globally frozen Issuance fails&lt;/li&gt;
&lt;li&gt;Test Convert zero amount from unauthorized account fails&lt;/li&gt;
&lt;li&gt;Test Holder unauthorize with nonzero confidential balance fails&lt;/li&gt;
&lt;li&gt;Test Issuance set rotate existing issuer encryption key fails&lt;/li&gt;
&lt;li&gt;Test Issuance set add auditor after holders have confidential balance fails&lt;/li&gt;
&lt;li&gt;Test Create with privacy locked off then enable attempt fails&lt;/li&gt;
&lt;li&gt;Test Create with invalid mutable flag bit fails&lt;/li&gt;
&lt;li&gt;Test Midflight auditor cannot be retrofitted after issuer key already set&lt;/li&gt;
&lt;li&gt;Test Midflight set require auth after holders have cb freezes unallowlisted&lt;/li&gt;
&lt;li&gt;Test Midflight change domain id strands holder not in new domain&lt;/li&gt;
&lt;li&gt;Test Midflight per holder lock blocks holder unauthorize&lt;/li&gt;
&lt;li&gt;Test Crash send all zero zkproof&lt;/li&gt;
&lt;li&gt;Test Crash send all ones zkproof&lt;/li&gt;
&lt;li&gt;Test Crash send random garbage zkproof&lt;/li&gt;
&lt;li&gt;Test Crash send bit flipped zkproof&lt;/li&gt;
&lt;li&gt;Test Crash send zero amount commitment&lt;/li&gt;
&lt;li&gt;Test Crash send zero balance commitment&lt;/li&gt;
&lt;li&gt;Test Crash send zero sender encrypted amount&lt;/li&gt;
&lt;li&gt;Test Crash send swapped sender dest ciphertexts&lt;/li&gt;
&lt;li&gt;Test Crash convert all zero zkproof&lt;/li&gt;
&lt;li&gt;Test Crash convert back all zero zkproof&lt;/li&gt;
&lt;li&gt;Test Crash clawback all zero zkproof&lt;/li&gt;
&lt;li&gt;Test Crash send rapid fire garbage proofs no corruption&lt;/li&gt;
&lt;li&gt;Test Acctdel holder with active cb s blocks account delete&lt;/li&gt;
&lt;li&gt;Test Acctdel holder with unmerged inbox blocks account delete&lt;/li&gt;
&lt;li&gt;Test Acctdel issuer with active confidential outstanding blocks&lt;/li&gt;
&lt;li&gt;Test Acctdel orphan mptoken after Issuance destroy full cleanup chain&lt;/li&gt;
&lt;li&gt;Test Acctdel issuer after destroy with outstanding orphan holders succeeds&lt;/li&gt;
&lt;li&gt;Test Acctdel stuck holder after issuer unauthorize with cb present&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Adversarial / Security
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Adversarial duplicate send txn&lt;/li&gt;
&lt;li&gt;Test Adversarial stale send proof after merge&lt;/li&gt;
&lt;li&gt;Test Adversarial reuse convert proof after state change&lt;/li&gt;
&lt;li&gt;Test Adversarial reuse convert back proof after state change&lt;/li&gt;
&lt;li&gt;Test Adversarial two sends exceed balance same ledger&lt;/li&gt;
&lt;li&gt;Test Adversarial send and convert back same balance same ledger&lt;/li&gt;
&lt;li&gt;Test Adversarial send vs clawback race same ledger&lt;/li&gt;
&lt;li&gt;Test Adversarial valid ec point wrong value&lt;/li&gt;
&lt;li&gt;Test Adversarial send amount commitment aliased to balance commitment&lt;/li&gt;
&lt;li&gt;Test Adversarial swap holder and issuer encrypted amounts&lt;/li&gt;
&lt;li&gt;Test Adversarial zero byte zkproof&lt;/li&gt;
&lt;li&gt;Test Adversarial proof for different statement&lt;/li&gt;
&lt;li&gt;Test Adversarial point on wrong curve&lt;/li&gt;
&lt;li&gt;Test Adversarial identity element as ciphertext&lt;/li&gt;
&lt;li&gt;Test Adversarial all zero blinding factor&lt;/li&gt;
&lt;li&gt;Test Adversarial range proof negative modular&lt;/li&gt;
&lt;li&gt;Test Adversarial nonce reuse across sends&lt;/li&gt;
&lt;li&gt;Test Adversarial wrong holder key in send&lt;/li&gt;
&lt;li&gt;Test Adversarial holder amount encrypted under issuer key&lt;/li&gt;
&lt;li&gt;Test Adversarial issuer amount encrypted under holder key&lt;/li&gt;
&lt;li&gt;Test Adversarial valid proof wrong account signature&lt;/li&gt;
&lt;li&gt;Test Adversarial key hijacking first convert&lt;/li&gt;
&lt;li&gt;Test Adversarial auditor amount wrong key&lt;/li&gt;
&lt;li&gt;Test Adversarial auditor amount different value&lt;/li&gt;
&lt;li&gt;Test Adversarial stale auditor key after rotation&lt;/li&gt;
&lt;li&gt;Test Adversarial range proof near uint64 overflow&lt;/li&gt;
&lt;li&gt;Test Adversarial cb in amount tampered&lt;/li&gt;
&lt;li&gt;Test Adversarial convert back tampered amount exceeds proof&lt;/li&gt;
&lt;li&gt;Test Adversarial send from inbox without merge&lt;/li&gt;
&lt;li&gt;Test Adversarial convert then immediate convert back same ledger&lt;/li&gt;
&lt;li&gt;Test Adversarial clawback from destroyed mptoken&lt;/li&gt;
&lt;li&gt;Test Adversarial clawback after holder converts back to zero&lt;/li&gt;
&lt;li&gt;Test Adversarial send with extra unknown fields&lt;/li&gt;
&lt;li&gt;Test Adversarial send all required fields missing&lt;/li&gt;
&lt;li&gt;Test Adversarial holder encrypted amount wrong byte length&lt;/li&gt;
&lt;li&gt;Test Adversarial issuer encrypted amount truncated&lt;/li&gt;
&lt;li&gt;Test Adversarial zkproof truncated&lt;/li&gt;
&lt;li&gt;Test Adversarial zkproof padded with extra bytes&lt;/li&gt;
&lt;li&gt;Test Adversarial confidential state persists across operations&lt;/li&gt;
&lt;li&gt;Test Adversarial two sends canonical order&lt;/li&gt;
&lt;li&gt;Test Adversarial replay with new sequence&lt;/li&gt;
&lt;li&gt;Test Adversarial cross type zkp replay&lt;/li&gt;
&lt;li&gt;Test Adversarial non issuer clawback valid zkp&lt;/li&gt;
&lt;li&gt;Test Adversarial third party merge inbox valid zkp&lt;/li&gt;
&lt;li&gt;Test Adversarial enable privacy on existing transfer fee Issuance reverse mutex&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Invariants
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Inv send increments version when spending changes&lt;/li&gt;
&lt;li&gt;Test Inv convert creates consistent encrypted fields&lt;/li&gt;
&lt;li&gt;Test Inv send preserves encrypted field coexistence&lt;/li&gt;
&lt;li&gt;Test Inv convert token conservation holder and issuer side&lt;/li&gt;
&lt;li&gt;Test Inv send does not modify public mptamount at either end&lt;/li&gt;
&lt;li&gt;Test Inv send does not modify Issuance outstandingamount&lt;/li&gt;
&lt;li&gt;Test Inv clawback decreases coa and oa by identical amount&lt;/li&gt;
&lt;li&gt;Test Clawback under global lock no invariant&lt;/li&gt;
&lt;li&gt;Test Convert back under global lock returns teclocked&lt;/li&gt;
&lt;li&gt;Test Inv coa never exceeds oa through chain&lt;/li&gt;
&lt;li&gt;Test Inv send with tampered amount commitment rejected leaves coa unchanged&lt;/li&gt;
&lt;li&gt;Test Inv full clawback zeros holder cb and removes their coa contribution&lt;/li&gt;
&lt;li&gt;Test Inv full convert back chain leaves no orphan confidential state&lt;/li&gt;
&lt;li&gt;Test Convert with zero amount asymmetric behavior probe&lt;/li&gt;
&lt;li&gt;Test Convert back with zero amount returns temBAD AMOUNT&lt;/li&gt;
&lt;li&gt;Test Clawback with zero amount returns temBAD AMOUNT&lt;/li&gt;
&lt;li&gt;Test Send to self returns temMALFORMED&lt;/li&gt;
&lt;li&gt;Test Convert with amount exceeding public balance returns tecINSUFFICIENT FUNDS&lt;/li&gt;
&lt;li&gt;Test Two issuances independent version counters&lt;/li&gt;
&lt;li&gt;Test Two issuances independent coa accounting&lt;/li&gt;
&lt;li&gt;Test Two issuances independent auditor decryption&lt;/li&gt;
&lt;li&gt;Test Account delete blocked when alice has cb on one of two issuances&lt;/li&gt;
&lt;li&gt;Test Convert back on Issuance a does not affect balance on b&lt;/li&gt;
&lt;li&gt;Test Clawback on Issuance a does not affect balance on b&lt;/li&gt;
&lt;li&gt;Test scenario A send with auditor all zero zkproof rejected&lt;/li&gt;
&lt;li&gt;Test scenario A send with auditor random garbage zkproof rejected&lt;/li&gt;
&lt;li&gt;Test scenario A send with auditor bit flipped zkproof rejected&lt;/li&gt;
&lt;li&gt;Test scenario A send with auditor truncated zkproof rejected&lt;/li&gt;
&lt;li&gt;Test scenario B convert MPT amount as string above 2 pow 32 succeeds&lt;/li&gt;
&lt;li&gt;Test scenario B convert MPT amount as int above 2 pow 32 is rejected&lt;/li&gt;
&lt;li&gt;Test scenario B convert two holders summing above 2 pow 32 keeps coa correct&lt;/li&gt;
&lt;li&gt;Test scenario B Issuance create with maximum amount at 2 pow 63 minus 1&lt;/li&gt;
&lt;li&gt;Test scenario B Issuance create with maximum amount above 2 pow 63 is rejected&lt;/li&gt;
&lt;li&gt;Test scenario C convert back from inbox only no spending is rejected&lt;/li&gt;
&lt;li&gt;Test scenario C send from inbox only no spending is rejected&lt;/li&gt;
&lt;li&gt;Test scenario C merge inbox after inbox only state promotes value to spending&lt;/li&gt;
&lt;li&gt;Test scenario D send after issuer unauthorizes sender fails&lt;/li&gt;
&lt;li&gt;Test scenario E clawback after issuer unauthorizes holder succeeds&lt;/li&gt;
&lt;li&gt;Test scenario F three holder full chain conserves total value&lt;/li&gt;
&lt;li&gt;Test scenario F two Issuance concurrent chains remain isolated&lt;/li&gt;
&lt;li&gt;Test scenario H zero amount convert registers holder key and enables inbound send&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Batch / Composition
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch confidential MPT two sends with stale proof allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT independent send with one failing&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT convert then merge inbox allornothing&lt;/li&gt;
&lt;li&gt;Test Batch delegated send missing permission allornothing&lt;/li&gt;
&lt;li&gt;Test Batch delegated send with delegate as outer needs no batch signer&lt;/li&gt;
&lt;li&gt;Test Batch delegated send with delegate as outer with batch signer&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT two chained sends same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT send then convert back same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT three chained sends same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential MPT chained send wrong predicted ciphertext fails&lt;/li&gt;
&lt;li&gt;Test Batch only one with two valid chained sends executes first only&lt;/li&gt;
&lt;li&gt;Test Batch only one first send to unconverted fails falls through to second&lt;/li&gt;
&lt;li&gt;Test Batch until failure first send commits then stale second fails&lt;/li&gt;
&lt;li&gt;Test Batch allornothing issuer lock then alice send rolls back both&lt;/li&gt;
&lt;li&gt;Test Batch independent three holders converts coa aggregate correct&lt;/li&gt;
&lt;li&gt;Test Batch allornothing inner revokes delegate then later inner uses it rolls back&lt;/li&gt;
&lt;li&gt;Test Batch two chained delegated sends from same delegator both commit&lt;/li&gt;
&lt;li&gt;Test Delegated send after delegator loses domain membership fails&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Delegation
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Delegate confidential MPT send basic&lt;/li&gt;
&lt;li&gt;Test Delegate confidential MPT send missing permission&lt;/li&gt;
&lt;li&gt;Test Delegate confidential MPT convert&lt;/li&gt;
&lt;li&gt;Test Delegate confidential MPT merge inbox&lt;/li&gt;
&lt;li&gt;Test Delegate confidential MPT convert back&lt;/li&gt;
&lt;li&gt;Test Delegate confidential MPT clawback by issuer&lt;/li&gt;
&lt;li&gt;Test Delegate neg send perm only cannot convert back&lt;/li&gt;
&lt;li&gt;Test Delegate neg convert back perm only cannot send&lt;/li&gt;
&lt;li&gt;Test Delegate neg clawback with non issuer account rejected at preflight&lt;/li&gt;
&lt;li&gt;Test Delegate neg revoked permission blocks subsequent use&lt;/li&gt;
&lt;li&gt;Test Delegate neg self delegation rejected&lt;/li&gt;
&lt;li&gt;Test Delegate neg send with proof over delegate keys rejected&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Cross-Feature
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test Ticket based first convert succeeds when proof binds to ticket seq&lt;/li&gt;
&lt;li&gt;Test Ticket based send succeeds when proof binds to ticket seq&lt;/li&gt;
&lt;li&gt;Test Ticket based send with wrong seq in proof returns tecBAD PROOF&lt;/li&gt;
&lt;li&gt;Test Ticket based convert back succeeds when proof binds to ticket seq&lt;/li&gt;
&lt;li&gt;Test Ticket based send with nonzero sequence returns temSEQ AND TICKET&lt;/li&gt;
&lt;li&gt;Test Ticket based send with nonexistent ticket is rejected&lt;/li&gt;
&lt;li&gt;Test Ticket based send consumes ticket even on tecBAD PROOF&lt;/li&gt;
&lt;li&gt;Test Ticket based send with matching credential succeeds&lt;/li&gt;
&lt;li&gt;Test Escrow create with MPT locks only public balance&lt;/li&gt;
&lt;li&gt;Test Escrow finish credits destination public not CB&lt;/li&gt;
&lt;li&gt;Test Confidential send succeeds with active outgoing escrow&lt;/li&gt;
&lt;li&gt;Test Escrow create amount exceeding unlocked public returns tecINSUFFICIENT FUNDS&lt;/li&gt;
&lt;li&gt;Test Check create then cash on privacy MPT does not touch either side CB&lt;/li&gt;
&lt;li&gt;Test Convert amount capped by unlocked public not total owned&lt;/li&gt;
&lt;li&gt;Test Vault create with confidential enabled MPT succeeds&lt;/li&gt;
&lt;li&gt;Test Vault deposit preserves holder confidential state&lt;/li&gt;
&lt;li&gt;Test Vault deposit fails when holder has no public balance&lt;/li&gt;
&lt;li&gt;Test Vault withdraw preserves destination confidential state&lt;/li&gt;
&lt;li&gt;Test Vault pseudo mptoken has no confidential fields after deposit&lt;/li&gt;
&lt;li&gt;Test Vault confidential clawback independent of vault holdings&lt;/li&gt;
&lt;li&gt;Test Send at exactly 10x base fee succeeds&lt;/li&gt;
&lt;li&gt;Test Send at one drop below 10x base fee returns telINSUF FEE P&lt;/li&gt;
&lt;li&gt;Test First convert does not change holder owner count&lt;/li&gt;
&lt;li&gt;Test Issuance create then key register bumps issuer owner count by exactly 1&lt;/li&gt;
&lt;li&gt;Test Five confidential sends drain alice xrp by exactly 5x send fee&lt;/li&gt;
&lt;li&gt;Test Memo on confidential send does not affect proof&lt;/li&gt;
&lt;li&gt;Test Multi sig confidential send at exactly required fee succeeds&lt;/li&gt;
&lt;li&gt;Test Multi sig confidential send at one drop below required returns telINSUF FEE P&lt;/li&gt;
&lt;li&gt;Test Convert does not update sfOutstandingAmount&lt;/li&gt;
&lt;li&gt;Test Convert fee multiplier 10x bounds&lt;/li&gt;
&lt;li&gt;Test Merge inbox fee multiplier 10x bounds&lt;/li&gt;
&lt;li&gt;Test Convert back fee multiplier 10x bounds&lt;/li&gt;
&lt;li&gt;Test Clawback fee multiplier 10x bounds&lt;/li&gt;
&lt;li&gt;Test Send to deposit auth destination without preauth fails&lt;/li&gt;
&lt;li&gt;Test Send to deposit auth destination after direct preauth succeeds&lt;/li&gt;
&lt;li&gt;Test Send to deposit auth destination after unauth fails again&lt;/li&gt;
&lt;li&gt;Test Send to deposit auth destination with matching credential succeeds&lt;/li&gt;
&lt;li&gt;Test Send after credential deleted fails with nonexistent credential&lt;/li&gt;
&lt;li&gt;Test Send with third party credential id fails&lt;/li&gt;
&lt;li&gt;Test Send with unaccepted credential fails&lt;/li&gt;
&lt;li&gt;Test Send with duplicate credential ids fails&lt;/li&gt;
&lt;li&gt;Test Send with credential ids when destination has no deposit auth succeeds&lt;/li&gt;
&lt;li&gt;Test Credential based send with frozen destination fails with lock&lt;/li&gt;
&lt;li&gt;Test Failed credential based send does not credit destination inbox&lt;/li&gt;
&lt;li&gt;Test Credential based send credits all encrypted amounts when auditor present&lt;/li&gt;
&lt;li&gt;Test Issuance set domain id on Issuance without require auth fails&lt;/li&gt;
&lt;li&gt;Test Issuance set nonexistent domain id fails&lt;/li&gt;
&lt;li&gt;Test Confidential send in permissioned domain succeeds&lt;/li&gt;
&lt;li&gt;Test Pd payment to uncredentialed holder blocks funding&lt;/li&gt;
&lt;li&gt;Test Pd send when destination credential revoked blocks send&lt;/li&gt;
&lt;li&gt;Test Pd send when sender credential revoked blocks send&lt;/li&gt;
&lt;li&gt;Test Pd send after domain id added post convert succeeds for credentialed&lt;/li&gt;
&lt;li&gt;Test Pd clear domain id via zero sentinel revokes holder authorization&lt;/li&gt;
&lt;li&gt;Test Pd send with unaccepted destination credential blocks send&lt;/li&gt;
&lt;li&gt;Test Convert with reserve sponsor returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Merge inbox with reserve sponsor returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Send with reserve sponsor returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Convert back with reserve sponsor returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Clawback with reserve sponsor returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Send with fee plus reserve flags returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Send with zero sponsor flags returns temINVALID FLAG&lt;/li&gt;
&lt;li&gt;Test Send with sponsor equal account returns temMALFORMED&lt;/li&gt;
&lt;li&gt;Test Prefunded fee sponsorship for confidential send succeeds&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Bug Hunt
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Test sum IssuerEncryptedBalance equals coa after multi hop&lt;/li&gt;
&lt;li&gt;Test IssuerEncryptedBalance zero after full clawback&lt;/li&gt;
&lt;li&gt;Test IssuerEncryptedBalance decrypts to expected after partial send&lt;/li&gt;
&lt;li&gt;Test stale version send rejected after intervening send&lt;/li&gt;
&lt;li&gt;Test convertback tampered MPT amount field rejected&lt;/li&gt;
&lt;li&gt;Test send balance commitment all zero bytes rejected&lt;/li&gt;
&lt;li&gt;Test send amount commitment all zero bytes rejected&lt;/li&gt;
&lt;li&gt;Test holder key immutable second convert with new key rejected&lt;/li&gt;
&lt;li&gt;Test convert with holder key equal to issuer key rejected&lt;/li&gt;
&lt;li&gt;Test first convert initializes spending to enczero and version zero&lt;/li&gt;
&lt;li&gt;Test second convert does not mutate spending or version&lt;/li&gt;
&lt;li&gt;Test issuer can re encrypt holder balance under new auditor key off chain&lt;/li&gt;
&lt;li&gt;Test enable auditor after convert send should not tefinternal&lt;/li&gt;
&lt;li&gt;Test mergeinbox on encrypted zero inbox version semantics&lt;/li&gt;
&lt;li&gt;Test stale send proof after clawback rejected&lt;/li&gt;
&lt;li&gt;Test zero amount convert registers key and accepts inbound&lt;/li&gt;
&lt;li&gt;Test send ciphertext under wrong key reaches bad proof not internal&lt;/li&gt;
&lt;li&gt;Test send swap ciphertexts from different amount proof rejected&lt;/li&gt;
&lt;li&gt;Test send to authorized but unconverted holder clean reject&lt;/li&gt;
&lt;li&gt;Test batch ct sends allornothing rolls back version on final failure&lt;/li&gt;
&lt;li&gt;Test freeze and send same ledger ordering is clean&lt;/li&gt;
&lt;li&gt;Test iss enc key reset after full clawback blocked pre reset&lt;/li&gt;
&lt;li&gt;Test convert with holder key not owned by account silent loss&lt;/li&gt;
&lt;li&gt;Test convert at coa boundary rejects cleanly not internal&lt;/li&gt;
&lt;li&gt;Test Add auditor after zero amount convert does not crash next convert&lt;/li&gt;
&lt;li&gt;Test Add auditor after convert convertback roundtrip blocked by gates&lt;/li&gt;
&lt;li&gt;Test Convert back amount within cb s plus cb in but exceeding cb s rejected&lt;/li&gt;
&lt;li&gt;Test Recipient merge between two sender sends does not break sender&lt;/li&gt;
&lt;li&gt;Test Clawback when holder has only unmerged inbox state reset correctly&lt;/li&gt;
&lt;/ol&gt;

</description>
      <category>automation</category>
      <category>blockchain</category>
      <category>crypto</category>
      <category>testing</category>
    </item>
    <item>
      <title>Sponsored Fees and Reserves — Performance Test Report</title>
      <dc:creator>Qi Zhao</dc:creator>
      <pubDate>Tue, 22 Sep 2026 15:07:34 +0000</pubDate>
      <link>https://dev.to/ripplexdev/sponsored-fees-and-reserves-performance-test-report-10da</link>
      <guid>https://dev.to/ripplexdev/sponsored-fees-and-reserves-performance-test-report-10da</guid>
      <description>&lt;p&gt;&lt;em&gt;By Hitesh Gandhi(&lt;a class="mentioned-user" href="https://dev.to/hgandhi"&gt;@hgandhi&lt;/a&gt;) and Qi Zhao&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;The &lt;code&gt;Sponsor&lt;/code&gt; amendment (&lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0068-sponsored-fees-and-reserves" rel="noopener noreferrer"&gt;XLS-68: Sponsored Fees and Reserves&lt;/a&gt;) introduces a way for one XRPL account to pay transaction fees and owner reserves on behalf of another account without taking custody of that account's XRP. This lets exchanges, wallets, token issuers, and applications simplify onboarding by covering network costs for their users while those users retain full control of their keys and accounts.&lt;/p&gt;

&lt;p&gt;XLS-68 supports two sponsorship modes. In &lt;strong&gt;co-signed&lt;/strong&gt; sponsorship, the sponsor's signature is attached to each transaction via the &lt;code&gt;Sponsor&lt;/code&gt; transaction field. In &lt;strong&gt;pre-funded&lt;/strong&gt; sponsorship, the sponsor creates a &lt;code&gt;Sponsorship&lt;/code&gt; ledger object holding a fee pool (&lt;code&gt;FeeAmount&lt;/code&gt;) and a per-transaction ceiling (&lt;code&gt;MaxFee&lt;/code&gt;), so the sponsee can transact without the sponsor co-signing every time. Two new transaction types are added — &lt;code&gt;SponsorshipSet&lt;/code&gt; and &lt;code&gt;SponsorshipTransfer&lt;/code&gt; — and sponsorship applies to both account and object reserves.&lt;/p&gt;

&lt;p&gt;The performance of XLS-68 was evaluated in a private XRPL performance network running production-representative infrastructure. Testing covered mixed-load simulations, stress testing at 2× baseline capacity, endurance runs of up to 5 hours, 2-signer multisign validation, and complex 6-path/8-step pathfinding under sponsorship load.&lt;/p&gt;




&lt;h2&gt;
  
  
  Executive Summary
&lt;/h2&gt;

&lt;p&gt;The &lt;code&gt;Sponsor&lt;/code&gt; amendment introduces no measurable performance degradation to XRPL consensus processing.&lt;/p&gt;

&lt;p&gt;Under the most demanding realistic workload — 12 transaction types including complex 6-path/8-step DEX pathfinding alongside all sponsorship operations — the network sustained &lt;strong&gt;322 TPS aggregate with zero overvalidations&lt;/strong&gt; in the 60-minute flagship run. Under 2× stress, the network delivered &lt;strong&gt;450 TPS sustained&lt;/strong&gt; with a peak burst of &lt;strong&gt;655 TPS&lt;/strong&gt; and zero overvalidations: a &lt;strong&gt;90.9% throughput increase for only a 6.6% increase in close rate&lt;/strong&gt;, demonstrating sublinear consensus overhead.&lt;/p&gt;

&lt;h3&gt;
  
  
  Key Performance Metrics at a Glance
&lt;/h3&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;th&gt;Significance&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Peak sustained throughput&lt;/td&gt;
&lt;td&gt;450 TPS&lt;/td&gt;
&lt;td&gt;2× baseline with zero overvalidations&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg consensus latency&lt;/td&gt;
&lt;td&gt;3.54–3.89 s&lt;/td&gt;
&lt;td&gt;Well within 5-second target&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P95 consensus latency&lt;/td&gt;
&lt;td&gt;3.95–4.27 s&lt;/td&gt;
&lt;td&gt;No tail-latency concerns&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidation rate&lt;/td&gt;
&lt;td&gt;≤ 0.4% per run&lt;/td&gt;
&lt;td&gt;Negligible across all test phases&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Validator out-of-sync events&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Perfect consensus alignment&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P2P node out-of-sync events&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Full network stability maintained&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Error rate (sponsorship TXs)&lt;/td&gt;
&lt;td&gt;≤ 0.23%&lt;/td&gt;
&lt;td&gt;Production-grade reliability&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The results show that XLS-68 is performant, stable, and ready for MainNet activation from a performance perspective.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Specification reference:&lt;/strong&gt; &lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0068-sponsored-fees-and-reserves" rel="noopener noreferrer"&gt;XLS-0068: Sponsored Fees and Reserves&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  Testing Objectives
&lt;/h2&gt;

&lt;p&gt;The primary objective was to validate that sponsored transaction processing does not degrade XRPL consensus health under sustained and burst loads. Specifically, we aimed to:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Validate that mixed sponsored and non-sponsored workloads maintain sub-5-second consensus latency under sustained load.&lt;/li&gt;
&lt;li&gt;Stress test at 2× baseline to identify capacity headroom and breaking points.&lt;/li&gt;
&lt;li&gt;Measure per-transaction-type throughput, latency, and error rates across all sponsorship modes (pre-funded, co-signed, multisign).&lt;/li&gt;
&lt;li&gt;Validate endurance stability, confirming no memory leaks, CPU drift, or progressive degradation.&lt;/li&gt;
&lt;li&gt;Confirm that sponsorship transactions coexist safely with the most computationally expensive XRPL operations (6-path/8-step DEX pathfinding).&lt;/li&gt;
&lt;/ul&gt;




&lt;h2&gt;
  
  
  Testing Methodology
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Capacity Planning
&lt;/h3&gt;

&lt;p&gt;The 5-second consensus latency limit serves as the benchmark for network capacity. Any ledger whose publishing latency exceeds this threshold is counted as an "overvalidation." The methodology targets a load level at which the network sustainably maintains sub-5-second close rates, then pushes to 2× that baseline to characterize stress behavior.&lt;/p&gt;

&lt;p&gt;Throughout this report, &lt;strong&gt;close rate&lt;/strong&gt; refers to ledger publishing latency (the interval at which validated ledgers are published), and &lt;strong&gt;TPS&lt;/strong&gt; refers to aggregate ledger throughput across all transaction types.&lt;/p&gt;

&lt;h3&gt;
  
  
  Test Environment
&lt;/h3&gt;

&lt;p&gt;Testing was conducted in a private XRPL environment with 9 nodes, mirroring Ripple's MainNet hardware specifications.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Network Setup&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;5 nodes function as validator nodes&lt;/li&gt;
&lt;li&gt;4 nodes serve as client P2P nodes, interfacing with load generators&lt;/li&gt;
&lt;li&gt;All nodes are hosted on AWS EC2 &lt;code&gt;z1d.2xlarge&lt;/code&gt; instances: 8 CPU cores, 64 GB RAM, 300 GB NVMe SSD&lt;/li&gt;
&lt;li&gt;All nodes operate within the same AWS region, interconnected via a shared LAN&lt;/li&gt;
&lt;li&gt;4 distributed load generators (ch0–ch3) submit transactions to the 4 P2P nodes&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;xrpld Configuration&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;The xrpld configuration was sourced from a Ripple MainNet validator, with modifications made only as required for the test environment.&lt;/p&gt;

&lt;h3&gt;
  
  
  Test Data Setup
&lt;/h3&gt;

&lt;p&gt;A dedicated data strategy was developed for XLS-68 testing. At a high level:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Account isolation:&lt;/strong&gt; Sponsor and sponsee account pools were provisioned separately from the baseline transaction accounts, so the two workloads could not interfere with one another.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Balance stability:&lt;/strong&gt; Payment destinations were arranged so that XRP balances remained stable across long-duration runs.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scale:&lt;/strong&gt; Roughly 100,000 sponsor/sponsee account pairs were distributed across the four load generators. Pre-funded sponsorship objects were provisioned with a fee pool and a per-transaction fee ceiling, and multisign scenarios used 8,000 master accounts with a 2-signer quorum, signed client-side.&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Test Phases and Progression
&lt;/h3&gt;

&lt;p&gt;Testing followed a progressive strategy designed to systematically validate every dimension of sponsorship performance:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Phase&lt;/th&gt;
&lt;th&gt;Objective&lt;/th&gt;
&lt;th&gt;Load Profile&lt;/th&gt;
&lt;th&gt;Duration&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Phase 1: Mixed Load&lt;/td&gt;
&lt;td&gt;Validate sponsored TX processing alongside baseline payments&lt;/td&gt;
&lt;td&gt;240 TPS (6 TX types)&lt;/td&gt;
&lt;td&gt;3 × 30 min&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Phase 2: Endurance&lt;/td&gt;
&lt;td&gt;Confirm stability under sustained load; detect memory leaks&lt;/td&gt;
&lt;td&gt;240 TPS constant&lt;/td&gt;
&lt;td&gt;2 × 4 hours&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Phase 3: Stress (2×)&lt;/td&gt;
&lt;td&gt;Identify breaking points by doubling baseline throughput&lt;/td&gt;
&lt;td&gt;480 TPS constant&lt;/td&gt;
&lt;td&gt;2 × 30 min&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Phase 4: Multisign&lt;/td&gt;
&lt;td&gt;Validate 2-signer multisign overhead and endurance&lt;/td&gt;
&lt;td&gt;80 TPS multisign&lt;/td&gt;
&lt;td&gt;2 × 30 min + 2 × 5 hr&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Phase 5: Full Mixed Load + 6-Path/8-Step&lt;/td&gt;
&lt;td&gt;Production-realistic simulation with complex pathfinding&lt;/td&gt;
&lt;td&gt;322 TPS (12 TX types incl. pathfinding)&lt;/td&gt;
&lt;td&gt;30 min + 30 min + 60 min&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Edge Cases&lt;/td&gt;
&lt;td&gt;1:N sponsorship, termination flows&lt;/td&gt;
&lt;td&gt;Variable&lt;/td&gt;
&lt;td&gt;On-demand&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h3&gt;
  
  
  Transaction Scenarios Under Test
&lt;/h3&gt;

&lt;p&gt;The sponsorship mixed-load model distributes transactions across the following types at 240 TPS aggregate:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Transaction Type&lt;/th&gt;
&lt;th&gt;Sponsorship Mode&lt;/th&gt;
&lt;th&gt;Target TPS&lt;/th&gt;
&lt;th&gt;Description&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Normal XRP Payment&lt;/td&gt;
&lt;td&gt;None (baseline)&lt;/td&gt;
&lt;td&gt;60&lt;/td&gt;
&lt;td&gt;Non-sponsored XRP-to-XRP payment&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Prefund Payment&lt;/td&gt;
&lt;td&gt;Pre-funded fee&lt;/td&gt;
&lt;td&gt;60&lt;/td&gt;
&lt;td&gt;Sponsored payment where sponsor pays fee via &lt;code&gt;Sponsorship&lt;/code&gt; pool&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cosigned Payment&lt;/td&gt;
&lt;td&gt;Co-signed&lt;/td&gt;
&lt;td&gt;40&lt;/td&gt;
&lt;td&gt;Sponsored payment with dual signature (sponsor + sponsee)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Sponsored OfferCreate&lt;/td&gt;
&lt;td&gt;Co-signed&lt;/td&gt;
&lt;td&gt;40&lt;/td&gt;
&lt;td&gt;DEX offer creation with sponsor covering reserve&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipSet + CheckCreate&lt;/td&gt;
&lt;td&gt;Pre-funded&lt;/td&gt;
&lt;td&gt;20&lt;/td&gt;
&lt;td&gt;Create sponsorship pool + sponsored check creation&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipTransfer (Account)&lt;/td&gt;
&lt;td&gt;Transfer&lt;/td&gt;
&lt;td&gt;20&lt;/td&gt;
&lt;td&gt;Transfer entire account sponsorship to a different sponsor&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipTransfer (Object)&lt;/td&gt;
&lt;td&gt;Transfer&lt;/td&gt;
&lt;td&gt;20&lt;/td&gt;
&lt;td&gt;Transfer individual object sponsorship&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipSet (Delete)&lt;/td&gt;
&lt;td&gt;Termination&lt;/td&gt;
&lt;td&gt;Variable&lt;/td&gt;
&lt;td&gt;Terminate sponsorship relationships&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The full mixed load with 6-path/8-step pathfinding extends this to &lt;strong&gt;12 transaction types at a ~334 TPS target&lt;/strong&gt; by adding IOU Direct, 1Path1Step, 3Path3Step, and DEX 6-path/8-step payments.&lt;/p&gt;




&lt;h2&gt;
  
  
  Detailed Performance Results
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Phase 1 — Mixed Load Testing
&lt;/h3&gt;

&lt;p&gt;Three 30-minute mixed load runs were conducted at 240 TPS aggregate, combining sponsored and non-sponsored transactions. All three runs demonstrated highly consistent throughput and latency.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Run 1&lt;/th&gt;
&lt;th&gt;Run 2&lt;/th&gt;
&lt;th&gt;Run 3&lt;/th&gt;
&lt;th&gt;Assessment&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Ledgers Processed&lt;/td&gt;
&lt;td&gt;511&lt;/td&gt;
&lt;td&gt;544&lt;/td&gt;
&lt;td&gt;545&lt;/td&gt;
&lt;td&gt;Consistent ledger production&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;431,247&lt;/td&gt;
&lt;td&gt;437,095&lt;/td&gt;
&lt;td&gt;436,915&lt;/td&gt;
&lt;td&gt;~437K per run&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;236.03&lt;/td&gt;
&lt;td&gt;236.54&lt;/td&gt;
&lt;td&gt;234.99&lt;/td&gt;
&lt;td&gt;Stable at target&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;345.00&lt;/td&gt;
&lt;td&gt;296.33&lt;/td&gt;
&lt;td&gt;299.56&lt;/td&gt;
&lt;td&gt;Healthy burst capacity&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.65&lt;/td&gt;
&lt;td&gt;3.60&lt;/td&gt;
&lt;td&gt;3.59&lt;/td&gt;
&lt;td&gt;Well under 5 s target&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Negligible&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Validator out-of-sync&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Perfect sync&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P2P out-of-sync&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Perfect sync&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Key takeaway:&lt;/strong&gt; Across the three runs, throughput held steady at approximately 235 TPS and consensus latency remained well below the 5-second target. The core sponsorship transaction mix integrates cleanly with baseline XRPL payment processing.&lt;/p&gt;

&lt;h3&gt;
  
  
  Phase 2 — Endurance Testing
&lt;/h3&gt;

&lt;p&gt;Two independent 4-hour endurance runs validated sustained performance and memory stability under continuous sponsorship load.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Endurance Run 1&lt;/th&gt;
&lt;th&gt;Endurance Run 2&lt;/th&gt;
&lt;th&gt;Delta&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Duration&lt;/td&gt;
&lt;td&gt;4 hours&lt;/td&gt;
&lt;td&gt;4 hours&lt;/td&gt;
&lt;td&gt;n/a&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Ledgers Processed&lt;/td&gt;
&lt;td&gt;4,041&lt;/td&gt;
&lt;td&gt;4,042&lt;/td&gt;
&lt;td&gt;+0.02%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;3,209,083&lt;/td&gt;
&lt;td&gt;3,241,171&lt;/td&gt;
&lt;td&gt;+1.0%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;223.19&lt;/td&gt;
&lt;td&gt;224.97&lt;/td&gt;
&lt;td&gt;+0.8%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;436.24&lt;/td&gt;
&lt;td&gt;329.83&lt;/td&gt;
&lt;td&gt;Run 1 spike during online delete&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.61&lt;/td&gt;
&lt;td&gt;3.62&lt;/td&gt;
&lt;td&gt;+0.3%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;Negligible&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Key takeaway:&lt;/strong&gt; The two endurance runs produced virtually identical results, confirming repeatability. The Max TPS spike in Run 1 (436 TPS) correlates with xrpld's periodic online deletion cycle, which temporarily frees resources.&lt;/p&gt;

&lt;h3&gt;
  
  
  Phase 3 — Stress Testing (2× Baseline)
&lt;/h3&gt;

&lt;p&gt;Load was doubled to 480 TPS aggregate to identify the network's capacity ceiling with sponsored transactions.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Baseline (240 TPS)&lt;/th&gt;
&lt;th&gt;Stress (480 TPS)&lt;/th&gt;
&lt;th&gt;Change&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Ledgers Processed&lt;/td&gt;
&lt;td&gt;511&lt;/td&gt;
&lt;td&gt;504&lt;/td&gt;
&lt;td&gt;−1.4%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;431,247&lt;/td&gt;
&lt;td&gt;841,494&lt;/td&gt;
&lt;td&gt;+95.1%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;236.03&lt;/td&gt;
&lt;td&gt;450.54&lt;/td&gt;
&lt;td&gt;+90.9%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;345.00&lt;/td&gt;
&lt;td&gt;655.45&lt;/td&gt;
&lt;td&gt;+90.0%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.65&lt;/td&gt;
&lt;td&gt;3.89&lt;/td&gt;
&lt;td&gt;+6.6%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Improved&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Validator out-of-sync&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Zero&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P2P out-of-sync&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;Zero&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Key takeaway:&lt;/strong&gt; Doubling the transaction load from 240 to 480 TPS resulted in only a 6.6% increase in average close rate (3.65 s → 3.89 s). The network processed 841K transactions in the stress window — nearly double the baseline — without degradation in consensus health.&lt;/p&gt;

&lt;h3&gt;
  
  
  Phase 4 — Multisign (2-Signer) Testing
&lt;/h3&gt;

&lt;p&gt;Multisign transactions add cryptographic overhead because multiple signatures must be verified. The multisign test used a 2-signer flow with client-side signing.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Run 1 (30 min)&lt;/th&gt;
&lt;th&gt;Run 2 (30 min)&lt;/th&gt;
&lt;th&gt;Endurance 1 (5 hr)&lt;/th&gt;
&lt;th&gt;Endurance 2 (5 hr)&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;146,880&lt;/td&gt;
&lt;td&gt;146,880&lt;/td&gt;
&lt;td&gt;1,442,880&lt;/td&gt;
&lt;td&gt;1,442,880&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;78.98&lt;/td&gt;
&lt;td&gt;78.40&lt;/td&gt;
&lt;td&gt;80.18&lt;/td&gt;
&lt;td&gt;80.17&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;103.97&lt;/td&gt;
&lt;td&gt;116.84&lt;/td&gt;
&lt;td&gt;305.43&lt;/td&gt;
&lt;td&gt;106.95&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.543&lt;/td&gt;
&lt;td&gt;3.570&lt;/td&gt;
&lt;td&gt;3.546&lt;/td&gt;
&lt;td&gt;3.543&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P95 Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.959&lt;/td&gt;
&lt;td&gt;3.947&lt;/td&gt;
&lt;td&gt;3.952&lt;/td&gt;
&lt;td&gt;3.954&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Key takeaway:&lt;/strong&gt; 2-signer multisign transactions achieved the targeted 80 TPS with the lowest consensus latency of any phase (avg 3.54 s, P95 3.95 s), sustained across two 5-hour endurance runs.&lt;/p&gt;

&lt;h3&gt;
  
  
  Phase 5 — Full Mixed Load with Complex Pathfinding
&lt;/h3&gt;

&lt;p&gt;This phase represents the most demanding and production-realistic test scenario, combining all sponsorship transaction types with complex cross-currency pathfinding.&lt;/p&gt;

&lt;h4&gt;
  
  
  Aggregate Results
&lt;/h4&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Run 1 (30 min)&lt;/th&gt;
&lt;th&gt;Run 2 (30 min)&lt;/th&gt;
&lt;th&gt;Run 3 (60 min)&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Ledgers Processed&lt;/td&gt;
&lt;td&gt;504&lt;/td&gt;
&lt;td&gt;509&lt;/td&gt;
&lt;td&gt;979&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;614,366&lt;/td&gt;
&lt;td&gt;614,201&lt;/td&gt;
&lt;td&gt;1,183,972&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;329&lt;/td&gt;
&lt;td&gt;330&lt;/td&gt;
&lt;td&gt;322&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;464&lt;/td&gt;
&lt;td&gt;419&lt;/td&gt;
&lt;td&gt;416&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate (s)&lt;/td&gt;
&lt;td&gt;3.894&lt;/td&gt;
&lt;td&gt;3.852&lt;/td&gt;
&lt;td&gt;3.842&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;P95 Close Rate (s)&lt;/td&gt;
&lt;td&gt;4.272&lt;/td&gt;
&lt;td&gt;4.215&lt;/td&gt;
&lt;td&gt;4.158&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h4&gt;
  
  
  Per-Transaction Throughput Breakdown (60-Minute Run)
&lt;/h4&gt;

&lt;p&gt;The 60-minute test (Run 3) provides the most comprehensive view of sustained per-transaction performance:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Transaction Type&lt;/th&gt;
&lt;th&gt;Samples&lt;/th&gt;
&lt;th&gt;TPS&lt;/th&gt;
&lt;th&gt;Avg Latency (ms)&lt;/th&gt;
&lt;th&gt;P95 (ms)&lt;/th&gt;
&lt;th&gt;P99 (ms)&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Prefund Payment&lt;/td&gt;
&lt;td&gt;218,156&lt;/td&gt;
&lt;td&gt;59.60&lt;/td&gt;
&lt;td&gt;19.81&lt;/td&gt;
&lt;td&gt;79&lt;/td&gt;
&lt;td&gt;190&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cosigned Payment&lt;/td&gt;
&lt;td&gt;145,440&lt;/td&gt;
&lt;td&gt;39.74&lt;/td&gt;
&lt;td&gt;20.43&lt;/td&gt;
&lt;td&gt;71&lt;/td&gt;
&lt;td&gt;186&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Sponsored OfferCreate&lt;/td&gt;
&lt;td&gt;145,440&lt;/td&gt;
&lt;td&gt;39.74&lt;/td&gt;
&lt;td&gt;22.21&lt;/td&gt;
&lt;td&gt;85&lt;/td&gt;
&lt;td&gt;196&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipSet + CheckCreate&lt;/td&gt;
&lt;td&gt;33,332&lt;/td&gt;
&lt;td&gt;19.72&lt;/td&gt;
&lt;td&gt;32.87&lt;/td&gt;
&lt;td&gt;122&lt;/td&gt;
&lt;td&gt;297&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipTransfer (Account)&lt;/td&gt;
&lt;td&gt;72,720&lt;/td&gt;
&lt;td&gt;19.87&lt;/td&gt;
&lt;td&gt;18.79&lt;/td&gt;
&lt;td&gt;74&lt;/td&gt;
&lt;td&gt;201&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipTransfer (Object)&lt;/td&gt;
&lt;td&gt;72,720&lt;/td&gt;
&lt;td&gt;19.87&lt;/td&gt;
&lt;td&gt;18.44&lt;/td&gt;
&lt;td&gt;74&lt;/td&gt;
&lt;td&gt;193&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;SponsorshipSet (Delete)&lt;/td&gt;
&lt;td&gt;9,321&lt;/td&gt;
&lt;td&gt;2.59&lt;/td&gt;
&lt;td&gt;14.62&lt;/td&gt;
&lt;td&gt;76&lt;/td&gt;
&lt;td&gt;313&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IOU Direct Payment&lt;/td&gt;
&lt;td&gt;109,078&lt;/td&gt;
&lt;td&gt;29.80&lt;/td&gt;
&lt;td&gt;18.30&lt;/td&gt;
&lt;td&gt;71&lt;/td&gt;
&lt;td&gt;178&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;1Path1Step&lt;/td&gt;
&lt;td&gt;72,720&lt;/td&gt;
&lt;td&gt;19.87&lt;/td&gt;
&lt;td&gt;18.64&lt;/td&gt;
&lt;td&gt;75&lt;/td&gt;
&lt;td&gt;186&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3Path3Step&lt;/td&gt;
&lt;td&gt;72,720&lt;/td&gt;
&lt;td&gt;19.87&lt;/td&gt;
&lt;td&gt;22.46&lt;/td&gt;
&lt;td&gt;82&lt;/td&gt;
&lt;td&gt;186&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;DEX 6-Path/8-Step&lt;/td&gt;
&lt;td&gt;14,540&lt;/td&gt;
&lt;td&gt;3.97&lt;/td&gt;
&lt;td&gt;16.59&lt;/td&gt;
&lt;td&gt;53&lt;/td&gt;
&lt;td&gt;184&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Key takeaway:&lt;/strong&gt; The 60-minute mixed load run processed transactions at 322 TPS aggregate with zero overvalidations. Even the 6-path/8-step DEX payments — the most computationally expensive operation in the mix — held an average latency of 16.59 ms with a P99 of 184 ms.&lt;/p&gt;

&lt;h3&gt;
  
  
  Edge Case Testing
&lt;/h3&gt;

&lt;h4&gt;
  
  
  One-to-Many (1:N) Sponsorship
&lt;/h4&gt;

&lt;p&gt;Validated a single sponsor account funding multiple sponsees concurrently:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Pattern&lt;/td&gt;
&lt;td&gt;Single sponsor → multiple sponsees&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Total Transactions&lt;/td&gt;
&lt;td&gt;367,200&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg TPS&lt;/td&gt;
&lt;td&gt;198.11&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Max TPS&lt;/td&gt;
&lt;td&gt;250.40&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Avg Close Rate&lt;/td&gt;
&lt;td&gt;3.61 s&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Overvalidations&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h4&gt;
  
  
  Sponsorship Termination
&lt;/h4&gt;

&lt;p&gt;Bulk deletion of sponsorship relationships:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Metric&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Sponsorships Deleted&lt;/td&gt;
&lt;td&gt;40,000&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;TPS&lt;/td&gt;
&lt;td&gt;229.05&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Observed Errors&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;tefPAST_SEQ&lt;/code&gt;, &lt;code&gt;tecNO_ENTRY&lt;/code&gt; (expected under concurrent deletion)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;




&lt;h2&gt;
  
  
  System Resource Utilization
&lt;/h2&gt;

&lt;p&gt;System-level metrics were monitored continuously via Grafana dashboards across all test phases. The following analysis covers CPU, memory, disk I/O, and network utilization patterns.&lt;/p&gt;

&lt;h3&gt;
  
  
  CPU Utilization
&lt;/h3&gt;

&lt;p&gt;CPU utilization on validator nodes tracked linearly with TPS load across all test phases. During the 2× stress test (450 TPS achieved), CPU utilization increased proportionally without approaching saturation. Refer to the Grafana charts below for detailed CPU utilization trends.&lt;/p&gt;

&lt;h3&gt;
  
  
  Memory Utilization
&lt;/h3&gt;

&lt;p&gt;Memory utilization remained flat throughout all test phases, with no indication of memory leaks or unbounded growth in the xrpld process. This was confirmed across both endurance rounds. Memory growth correlated with expected ledger object expansion (new &lt;code&gt;Sponsorship&lt;/code&gt; objects, accounts, and offers) and was consistent with patterns observed in prior XRPL feature performance tests. Refer to the Grafana charts below for detailed memory utilization trends.&lt;/p&gt;

&lt;h3&gt;
  
  
  Disk I/O
&lt;/h3&gt;

&lt;p&gt;Disk I/O patterns showed normal activity with periodic spikes correlating to xrpld's online delete cycles and log rotation. One notable observation from the endurance tests: a transient CPU spike was traced to &lt;code&gt;logrotate&lt;/code&gt; compressing xrpld logs, which momentarily saturated disk I/O. This is an operational artifact rather than a feature concern, and is mitigated by scheduling log rotation during off-peak hours.&lt;/p&gt;

&lt;h3&gt;
  
  
  Network I/O
&lt;/h3&gt;

&lt;p&gt;Network utilization remained stable throughout all test phases, with no congestion or packet-loss events. Transaction propagation between P2P nodes and validators showed consistent sub-millisecond latency within the same-region LAN topology.&lt;/p&gt;

&lt;h3&gt;
  
  
  System Utilization Charts (60-Minute Mixed Load)
&lt;/h3&gt;

&lt;h4&gt;
  
  
  CPU
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8pdpaxeg4v4yka9gxznk.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F8pdpaxeg4v4yka9gxznk.png" alt="CPU utilization on validator nodes during the 60-minute mixed load run" width="800" height="134"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h4&gt;
  
  
  Memory
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwevhcpjea2owq9gnf013.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fwevhcpjea2owq9gnf013.png" alt="Memory utilization on validator nodes during the 60-minute mixed load run" width="799" height="262"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h4&gt;
  
  
  Disk I/O
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo1seo3lgtlxz4h2p87rf.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fo1seo3lgtlxz4h2p87rf.png" alt="Disk read and write activity on validator nodes during the 60-minute mixed load run" width="799" height="294"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h4&gt;
  
  
  Network
&lt;/h4&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Funm3f94boid55hwrfqbt.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Funm3f94boid55hwrfqbt.png" alt="Network throughput on validator nodes during the 60-minute mixed load run" width="799" height="357"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;The comprehensive performance evaluation of Sponsored Fees and Reserves (XLS-68) demonstrates that the amendment is ready, delivering high throughput, stable consensus, and efficient resource utilization across all tested scenarios.&lt;/p&gt;

&lt;h3&gt;
  
  
  Summary of Findings
&lt;/h3&gt;

&lt;p&gt;&lt;strong&gt;Throughput.&lt;/strong&gt; Sponsored transactions achieved 235–450 TPS in mixed load configurations, with a peak burst of 655 TPS under 2× stress.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Consensus stability.&lt;/strong&gt; Average consensus latency ranged from 3.54 s to 3.89 s across all test phases, consistently below the 5-second target. Overvalidations never exceeded 2 ledgers in any single run.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Endurance.&lt;/strong&gt; Extended runs of 4 and 5 hours confirmed sustained performance with zero validator or P2P sync issues and predictable, bounded memory growth.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Complex pathfinding coexistence.&lt;/strong&gt; The full mixed load with 6-path/8-step pathfinding — the most computationally demanding realistic scenario — achieved 322 TPS with zero overvalidations in the 60-minute run.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Multisign viability.&lt;/strong&gt; 2-signer multisign transactions sustained 80 TPS with sub-4-second consensus latency across two 5-hour endurance runs, processing 2.88 million transactions in total.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Resource efficiency.&lt;/strong&gt; CPU utilization remained well within capacity across all test phases including 2× stress, memory growth was bounded with no leaks detected, and no disk or network bottlenecks were observed.&lt;/p&gt;

&lt;h3&gt;
  
  
  Recommendation
&lt;/h3&gt;

&lt;p&gt;Based on the results of this comprehensive performance evaluation, the Sponsored Fees and Reserves (XLS-68) amendment is recommended for MainNet activation. All test phases achieved PASS status, with the network demonstrating ample capacity, stable consensus, and production-grade reliability under diverse and demanding workload conditions.&lt;/p&gt;

</description>
      <category>xrpl</category>
      <category>blockchain</category>
      <category>web3</category>
    </item>
    <item>
      <title>The XRP Ledger AI Starter Kit, v1.1: Open Standards, Not an Island</title>
      <dc:creator>Ajit Kulkarni</dc:creator>
      <pubDate>Tue, 15 Sep 2026 19:12:13 +0000</pubDate>
      <link>https://dev.to/ripplexdev/the-xrp-ledger-ai-starter-kit-v11-open-standards-not-an-island-358k</link>
      <guid>https://dev.to/ripplexdev/the-xrp-ledger-ai-starter-kit-v11-open-standards-not-an-island-358k</guid>
      <description>&lt;p&gt;&lt;strong&gt;TL;DR&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;In mid-June, we launched the XRPL AI Starter Kit to help developers build agentic payment applications on the XRP Ledger.&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Today we're shipping v1.1, adding support for two emerging cross-chain agent standards: the &lt;a href="https://openwallet.sh/" rel="noopener noreferrer"&gt;Open Wallet Standard (OWS)&lt;/a&gt; and &lt;a href="https://stripe.com/blog/machine-payments-protocol" rel="noopener noreferrer"&gt;Tempo and Stripe's Machine Payments Protocol (MPP)&lt;/a&gt;, a new CLI, &lt;code&gt;xrpl-up&lt;/code&gt;, for fast local testing, and a Skill for XRPL DEX trading.&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;&lt;em&gt;The throughline: XRPL isn't building agentic payments in isolation. It's showing up in the standards the rest of the agent ecosystem is converging on, and giving developers reference implementations for each.&lt;/em&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;When we introduced the &lt;a href="https://xrpl.org/docs/agents/agentic-transactions" rel="noopener noreferrer"&gt;XRPL AI Starter Kit&lt;/a&gt; in June, the goal was narrow and deliberate: give developers the fastest path from zero to a confirmed agent payment on the XRP Ledger, starting with X402 support for XRP and RLUSD. The XRPL has seen &lt;a href="https://xrpl-ai.org/" rel="noopener noreferrer"&gt;5.6M X402&lt;/a&gt; transactions since its inception.&lt;/p&gt;

&lt;p&gt;Since then, the agentic payments landscape has continued to evolve, and a few standards have emerged as genuine points of convergence rather than one-off frameworks. Wallet management is consolidating around shared, local-first interfaces instead of every tool reinventing its own keystore. And machine-to-machine settlement is consolidating around HTTP 402-based protocols that let any agent pay any service without an account or a checkout flow.&lt;/p&gt;

&lt;p&gt;XRPL AI Starter Kit v1.1 is about showing up in both of those places, not just the one we started with.&lt;/p&gt;

&lt;h2&gt;
  
  
  What's New in v1.1
&lt;/h2&gt;

&lt;h3&gt;
  
  
  1. XRPL support for the Open Wallet Standard (OWS)
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://openwallet.sh/" rel="noopener noreferrer"&gt;OWS&lt;/a&gt; is an open specification for local-first, multi-chain wallet storage and agent access to a single encrypted vault and a single interface (CLI, MCP, SDK, or REST) for every chain and every tool an agent might use. Instead of an agent's keys scattered across a dozen tool-specific config files and environment variables, OWS keeps them in a single vault at &lt;code&gt;~/.ows/&lt;/code&gt;, decrypted only long enough to sign, then wiped from memory.&lt;/p&gt;

&lt;p&gt;With v1.1, XRPL is supported as a chain in OWS. A single &lt;code&gt;ows wallet create&lt;/code&gt; command now derives an XRPL address alongside addresses for every other chain OWS supports, and agents can request XRPL signatures through the same policy-gated &lt;code&gt;ows_sign&lt;/code&gt; interface they'd use for any other chain with no XRPL-specific key handling required.&lt;/p&gt;

&lt;p&gt;We're also shipping &lt;a href="https://xrpl.org/docs/agents/xrpl-agent-wallet-skill" rel="noopener noreferrer"&gt;docs&lt;/a&gt; that walk through creating an OWS-backed XRPL wallet, wiring it to an agent, and sending a signed payment, all without the agent ever touching a raw private key.&lt;/p&gt;

&lt;p&gt;Why this matters: Agent frameworks don't want to write custom key management code for each chain. By supporting OWS, XRPL becomes accessible through the same wallet layer that developers are already adopting across other ecosystems.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. XRPL support for Tempo's Machine Payments Protocol (MPP)
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://stripe.com/blog/machine-payments-protocol" rel="noopener noreferrer"&gt;MPP&lt;/a&gt;, co-authored by Tempo and Stripe, is an open standard for HTTP-native agent payments: a service returns a 402 status with a price, the agent authorizes payment, and the resource is delivered. No accounts, no checkout flows.&lt;/p&gt;

&lt;p&gt;XRPL is now a supported settlement rail for MPP, live today for two payment modes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Single (charge) payments in XRP, IOU, and MPT.&lt;/strong&gt; An agent hits a 402 challenge, the XRPL payment settles in 3–5 seconds with sub-cent fees, and the resource is delivered. Deterministic finality means the transaction either confirms or it doesn't, with no ambiguous pending state for the agent to poll or retry.
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Session payments via XRPL Payment Channels for XRP.&lt;/strong&gt; This is MPP's most powerful primitive: instead of settling every API call individually, an agent opens a payment channel once, streams cumulative signed vouchers off-chain as it consumes a service, and the provider settles the total in a single on-chain transaction at the end. XRPL Payment Channels were built for exactly this pattern, where vouchers can be signed at rates exceeding 100,000 per second, verified locally by the provider with no blockchain call required, and the whole session costs just two on-chain transactions: open and close.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Stablecoin support for session payments is next. XRPL is adding IOU and MPT support to payment channels (tracked as &lt;a href="https://github.com/XRPLF/XRPL-Standards/discussions/287" rel="noopener noreferrer"&gt;XLS-93d&lt;/a&gt;), so RLUSD-denominated single payments and streaming sessions will follow the same lock-stream-settle mechanics as XRP today.&lt;/p&gt;

&lt;p&gt;For developers, the new &lt;a href="https://github.com/ripple/xrpl-mpp-sdk" rel="noopener noreferrer"&gt;&lt;code&gt;xrpl-mpp-sdk&lt;/code&gt;&lt;/a&gt; handles the full 402 challenge-response flow, credential signing, and session management for both charge and channel modes, namely, XRP, IOUs, and MPTs are all supported today, with demos covering everything from a basic XRP charge to a Claude agent paying per-prompt over a payment channel.&lt;br&gt;&lt;br&gt;
The npm package can be found &lt;a href="http://npmjs.com/package/xrpl-mpp-sdk" rel="noopener noreferrer"&gt;here&lt;/a&gt;, and the &lt;a href="https://mpp.dev/payment-methods/xrpl" rel="noopener noreferrer"&gt;MPP documentation&lt;/a&gt; also shows XRPL support. The npm package is in beta, so we welcome your feedback.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. &lt;code&gt;xrpl-up&lt;/code&gt;: a local CLI for fast developer iteration
&lt;/h3&gt;

&lt;p&gt;Testing agentic payment flows against the public Testnet or Devnet means facing faucet rate limits, ~4-second consensus times, and a shared state you don't control. &lt;a href="https://github.com/ripple/xrpl-up" rel="noopener noreferrer"&gt;&lt;code&gt;xrpl-up&lt;/code&gt;&lt;/a&gt; fixes that with a local sandbox: a one-command xrpld node with instant ledger closes, pre-funded accounts, and a full CLI surface for payments, trust lines, escrows, channels, AMM pools, MPTs, and more.&lt;/p&gt;

&lt;p&gt;It also ships a Claude Code plugin, so a developer can spin up a sandbox, create an AMM pool, or check a balance by describing what they want in plain language rather than having to remember flag syntax.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;xrpl-up&lt;/code&gt; isn't specific to agentic payments; it's a general local testing tool and the fastest way to iterate on any of the flows above (OWS wallets, MPP charges and channels, X402) without touching a public network.&lt;/p&gt;

&lt;h3&gt;
  
  
  4. An XRPL DEX trading Skill
&lt;/h3&gt;

&lt;p&gt;This &lt;a href="https://xrpl.org/docs/agents/getting-started-with-xrpl-trading" rel="noopener noreferrer"&gt;skill&lt;/a&gt; enables autonomous permissionless DEX trading on the XRPL. &lt;/p&gt;

&lt;h2&gt;
  
  
  Why This Matters: XRPL Is Not on an Island
&lt;/h2&gt;

&lt;p&gt;The OWS and MPP additions are XRPL showing up within a standard someone else is driving, rather than asking developers to adopt something XRPL-specific. OWS is a chain-agnostic wallet infrastructure with contributors spanning PayPal, Circle, the Solana Foundation, and others. XRPL is one of the chains it now speaks natively. MPP is a rail-agnostic settlement standard co-authored by Tempo and Stripe, and XRPL is one of the rails. X402, which we launched in June, works the same way.&lt;/p&gt;

&lt;p&gt;That pattern matters more than any single integration. The agentic payments ecosystem is still deciding which standards win, and XRPL's bet is to be a first-class citizen in the ones that do by bringing deterministic finality, predictable low fees, and native primitives like Payment Channels that map cleanly onto what these protocols actually need, rather than requiring a smart contract to bolt the behavior on.&lt;/p&gt;

&lt;h2&gt;
  
  
  Start Building
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Wallets:&lt;/strong&gt; Follow the new OWS tutorial to create an XRPL-backed wallet and wire it into an agent.
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Payments:&lt;/strong&gt; Try the &lt;code&gt;xrpl-mpp-sdk&lt;/code&gt; demos. Start with the basic XRP charge, then the payment-channel session demo to see the streaming pattern end-to-end.
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Local testing:&lt;/strong&gt; Install &lt;code&gt;xrpl-up&lt;/code&gt;, run &lt;code&gt;xrpl-up start&lt;/code&gt;, and point your scripts at &lt;code&gt;ws://localhost:6006&lt;/code&gt; instead of a public faucet.
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Still on X402?&lt;/strong&gt; Nothing here changes it, and X402 support for XRP and RLUSD from the original launch is unaffected.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;As with V1, this is a milestone in the journey, and not the finish line. Tell us what you're building, what's working, and what standard we should be looking at next.&lt;/p&gt;

</description>
      <category>agents</category>
      <category>ai</category>
      <category>blockchain</category>
      <category>crypto</category>
    </item>
    <item>
      <title>Batch V1.1: What Changed and Why It's Ready</title>
      <dc:creator>Mayukha Vadari</dc:creator>
      <pubDate>Mon, 14 Sep 2026 13:52:53 +0000</pubDate>
      <link>https://dev.to/ripplexdev/batch-v11-what-changed-and-why-its-ready-4ek0</link>
      <guid>https://dev.to/ripplexdev/batch-v11-what-changed-and-why-its-ready-4ek0</guid>
      <description>&lt;h1&gt;
  
  
  What is Batch?
&lt;/h1&gt;

&lt;p&gt;Batch (XLS-56) allows multiple XRPL transactions from different accounts to execute atomically in a single ledger close. For example, if any inner transaction fails during the all-or-nothing mode of the feature, the entire batch reverts. This makes various types of atomic operations much easier, and is also foundational infrastructure for multi-party coordination on the XRPL: atomic swaps, coordinated settlements, and any workflow where two or more parties need to move in lockstep without trust assumptions.&lt;/p&gt;

&lt;h1&gt;
  
  
  What happened with the original Batch (v1.0)
&lt;/h1&gt;

&lt;p&gt;On February 19, 2026, security researchers and &lt;a href="https://cantina.xyz/welcome" rel="noopener noreferrer"&gt;Cantina AI&lt;/a&gt; identified a critical logic flaw in the signature-validation routine of the original Batch amendment. The function &lt;code&gt;checkBatchSign&lt;/code&gt;, which validates that each inner-transaction account has authorized the batch, contained an early-return bug: when it encountered a signer whose account did not yet exist on the ledger, it immediately returned success and skipped validation of all remaining signers. This would have allowed an attacker to essentially execute arbitrary transactions on behalf of any victim accounts without their private keys.&lt;/p&gt;

&lt;p&gt;See the full Vulnerability Disclosure Report &lt;a href="https://xrpl.org/blog/2026/vulnerabilitydisclosurereport-bug-feb2026" rel="noopener noreferrer"&gt;here&lt;/a&gt;. &lt;/p&gt;

&lt;p&gt;The amendment was in its voting phase and had not been activated on Mainnet. No funds were at risk. Within hours of receiving the report, UNL validators were advised to vote "No," Ripple's engineering team reproduced the issue with an independent unit test, and an emergency release was published, marking both &lt;code&gt;Batch&lt;/code&gt; and &lt;code&gt;fixBatchInnerSigs&lt;/code&gt; as "unsupported".&lt;/p&gt;

&lt;h1&gt;
  
  
  What changed in Batch V1.1
&lt;/h1&gt;

&lt;p&gt;Batch V1.1 is a replacement amendment that ships the corrected implementation in &lt;code&gt;xrpld&lt;/code&gt; v3.3.0. The primary focus was on fixing the bug found in February and further hardening the implementation. &lt;/p&gt;

&lt;p&gt;We took the following steps to ensure that the new version is secure:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Checkpoint&lt;/th&gt;
&lt;th&gt;Status&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Root-cause fix (early-return elimination)&lt;/td&gt;
&lt;td&gt;Complete, merged via &lt;a href="https://github.com/XRPLF/rippled/pull/6446" rel="noopener noreferrer"&gt;PR #6446&lt;/a&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Code review by 4 senior engineers&lt;/td&gt;
&lt;td&gt;Complete&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;a href="https://audits.sherlock.xyz/contests/1260" rel="noopener noreferrer"&gt;Sherlock Batch Attackathon contest&lt;/a&gt;&lt;/td&gt;
&lt;td&gt;Complete&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;a href="https://www.halborn.com/audits/ripple/batch-re-audit-314089" rel="noopener noreferrer"&gt;Halborn Batch re-assessment&lt;/a&gt;&lt;/td&gt;
&lt;td&gt;Complete&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;a href="https://www.commonprefix.com/blog/xrp-ledger-batch-transactions-security-audit-report" rel="noopener noreferrer"&gt;Common Prefix Batch audit&lt;/a&gt;&lt;/td&gt;
&lt;td&gt;Complete&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI-assisted static analysis pipeline (including a Cantina scan)&lt;/td&gt;
&lt;td&gt;Integrated into review and release process&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;a href="https://dev.to/ripplexdev/batch-transaction-qa-test-report-5g5g"&gt;QA Devnet and testnet regression testing&lt;/a&gt;&lt;/td&gt;
&lt;td&gt;Complete&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;In addition, we found and fixed the following bugs:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Batch Transaction Bypasses MPT Validation, Crashes Node
&lt;/li&gt;
&lt;li&gt;Simulate RPC &lt;code&gt;tfInnerBatchTxn&lt;/code&gt; Flag Assertion Crash
&lt;/li&gt;
&lt;li&gt;Path Size Validation Bypass via Batch Transactions
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;preflight2&lt;/code&gt; &lt;code&gt;tfInnerBatchTxn&lt;/code&gt; flag bypasses signature verification in release builds
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;preflight1&lt;/code&gt; batch &lt;code&gt;parentBatchId&lt;/code&gt; invariant — assert-only enforcement with boolean logic error
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;checkValidity&lt;/code&gt; does not return early for valid inner batch transactions
&lt;/li&gt;
&lt;li&gt;Batch Signatures Miss Outer Account Binding
&lt;/li&gt;
&lt;li&gt;Batch signer ordering not enforced, allowing attacker to control authorization check order
&lt;/li&gt;
&lt;li&gt;Node crash via uncaught &lt;code&gt;std::runtime_error&lt;/code&gt; from &lt;code&gt;Batch::calculateBaseFee&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;buildBatchTxnIds&lt;/code&gt; hashes full inner array pre-count-cap at deserialization
&lt;/li&gt;
&lt;li&gt;Protocol bounds follow hashing and signature verification&lt;/li&gt;
&lt;/ul&gt;

&lt;h1&gt;
  
  
  Why Batch matters for the XRPL
&lt;/h1&gt;

&lt;p&gt;Batch is a prerequisite or accelerant for several high-value XRPL use cases already under contract or in active pipeline, such as:&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Atomic multi-party settlements.&lt;/strong&gt; Institutional workflows (custody transfers, DvP settlement, cross-border payment finalization) require two or more accounts to act simultaneously. Batch removes the need for trust in execution ordering, eliminating a class of race conditions and partial-execution risks.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Developer experience and ecosystem growth.&lt;/strong&gt; Batch is one of the most requested features by XRPL application developers and the community broadly. It enables patterns common on programmatic chains (such as multi-call, atomic bundles, and flash loans) natively on XRPL without smart contracts, reducing friction for new builders entering the ecosystem. Batch also enables DEXs, Wallets and Marketplaces to monetise on transactions they process by bundling the base transaction (such as a Payment) with a platform fee. There are many members of the XRPL ecosystem eager to use this feature.&lt;/p&gt;

&lt;h1&gt;
  
  
  Governance context
&lt;/h1&gt;

&lt;p&gt;The original Batch incident demonstrated that the XRPL governance model works. A critical bug was reported, validators responded within hours, and no funds were ever at risk because the amendment had not yet been activated. The v1.1 replacement was given additional scrutiny precisely because of that history: broader code review, AI-assisted auditing, Sherlock contest, a Halborn re-assessment, and a Common Prefix audit on top of the standard internal audit process.&lt;/p&gt;

&lt;p&gt;For validators voting on Batch V1.1 in &lt;code&gt;xrpld&lt;/code&gt; version 3.3.0:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;The root cause is fully understood and publicly documented (vulnerability disclosure report published February 25, 2026).
&lt;/li&gt;
&lt;li&gt;The fix addresses not just the specific bug but the class of vulnerability (early-return in validation loops, signing payload binding).
&lt;/li&gt;
&lt;li&gt;The review process has been expanded with additional tooling and reviewers relative to v1.0.
&lt;/li&gt;
&lt;li&gt;The business case is concrete: named partners, contracted use cases, and protocol-level dependencies are waiting on this capability.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://xrpl.org/docs/references/protocol/transactions/types/batch" rel="noopener noreferrer"&gt;Try it on Devnet today&lt;/a&gt;!&lt;/p&gt;

</description>
      <category>web3</category>
      <category>blockchain</category>
      <category>xrp</category>
      <category>xrpl</category>
    </item>
    <item>
      <title>Batch Transaction - QA Test Report</title>
      <dc:creator>Ramkumar SG</dc:creator>
      <pubDate>Tue, 08 Sep 2026 22:39:39 +0000</pubDate>
      <link>https://dev.to/ripplexdev/batch-transaction-qa-test-report-5g5g</link>
      <guid>https://dev.to/ripplexdev/batch-transaction-qa-test-report-5g5g</guid>
      <description>&lt;p&gt;&lt;strong&gt;Test Report Date&lt;/strong&gt;: 9/8/2026&lt;br&gt;
&lt;strong&gt;Prepared By&lt;/strong&gt;: &lt;a href="https://github.com/sgramkumar" rel="noopener noreferrer"&gt;sgramkumar&lt;/a&gt;&lt;br&gt;
&lt;strong&gt;Environment&lt;/strong&gt;: GitLab CI Runner (nix-debian)&lt;br&gt;
&lt;strong&gt;Network&lt;/strong&gt;: xrpld devnet + private CI network&lt;/p&gt;
&lt;h2&gt;
  
  
  Overview
&lt;/h2&gt;

&lt;p&gt;This report presents the results of QA testing performed on Batch (XLS-56) Transactions across xrpld servers. Coverage targets the  BatchV1_1  amendment, which supersedes the original  Batch  amendment and tightened BatchSigner authorization and signing semantics.&lt;/p&gt;
&lt;h2&gt;
  
  
  1. Feature
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Feature Name:&lt;/strong&gt; Batch (Atomic) Transactions&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Description:&lt;/strong&gt; Prior to this feature, an account that needed several operations to succeed or fail together had no way to bind them atomically; each transaction settled independently, leaving the risk of partial completion. Batch allows an account to package up to 8 inner transactions into a single outer &lt;code&gt;Batch&lt;/code&gt; transaction that is processed as one atomic unit. Inner transactions execute under one of four modes — &lt;code&gt;AllOrNothing&lt;/code&gt;, &lt;code&gt;OnlyOne&lt;/code&gt;, &lt;code&gt;UntilFailure&lt;/code&gt;, or &lt;code&gt;Independent&lt;/code&gt; — and may be authorized across multiple accounts (BatchSigners), via delegated authority, via multi-signing, or sequenced by tickets. This unlocks reliable multi-step and multi-party workflows on the XRP Ledger.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Specification Reference:&lt;/strong&gt; &lt;a href="https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0056-batch" rel="noopener noreferrer"&gt;https://github.com/XRPLF/XRPL-Standards/tree/master/XLS-0056-batch&lt;/a&gt;&lt;/p&gt;
&lt;h2&gt;
  
  
  2. Test Scope
&lt;/h2&gt;

&lt;p&gt;This report covers the full Batch surface as it stands under the &lt;code&gt;BatchV1_1&lt;/code&gt; amendment (xrpld PRs &lt;a href="https://github.com/XRPLF/rippled/pull/5060" rel="noopener noreferrer"&gt;#5060&lt;/a&gt;, &lt;a href="https://github.com/XRPLF/rippled/pull/6446" rel="noopener noreferrer"&gt;#6446&lt;/a&gt;). Testing focused on ensuring that:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;All four execution modes (&lt;code&gt;AllOrNothing&lt;/code&gt;, &lt;code&gt;OnlyOne&lt;/code&gt;, &lt;code&gt;UntilFailure&lt;/code&gt;, &lt;code&gt;Independent&lt;/code&gt;) behave per the XLS-56 specification, including correct commit/rollback semantics&lt;/li&gt;
&lt;li&gt;The 2–8 inner-transaction bounds, fee calculation, and &lt;code&gt;tfInnerBatchTxn&lt;/code&gt; rules are enforced&lt;/li&gt;
&lt;li&gt;Multi-account and multi-signed BatchSigner authorization is validated, including canonical signer ordering and the hardened &lt;code&gt;BatchV1_1&lt;/code&gt; signing preimage&lt;/li&gt;
&lt;li&gt;Batch composes correctly with the full range of inner transaction types (Payments, Offers, Checks, Escrow, MPT, Vault/Loan operations, and more), and with the features that have special interaction rules at the batch authorization/sequencing layer — Permission Delegation (inner  Delegate  field), Sponsored Fees &amp;amp; Reserves (outer/inner  Sponsor  rules), Confidential MPT (in-batch proof staleness), and Tickets (inner  TicketSequence)&lt;/li&gt;
&lt;li&gt;Security-critical properties hold under adversarial input, including malformed BatchSigners, stale-proof inner transactions, replay, and structural/serialization attacks&lt;/li&gt;
&lt;li&gt;Inner-transaction metadata (&lt;code&gt;sfParentBatchID&lt;/code&gt;) and same-ledger inclusion are correctly reported&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;
  
  
  3. Types of Testing Conducted
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Testing Type&lt;/th&gt;
&lt;th&gt;Description&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Functional&lt;/td&gt;
&lt;td&gt;Verifying each execution mode, inner-transaction type, fee path, and RPC/metadata surface against the XLS-56 specification&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Regression&lt;/td&gt;
&lt;td&gt;Running the full xrpld test suite to confirm Batch changes did not break existing functionality&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Adversarial / Security&lt;/td&gt;
&lt;td&gt;Attack-matrix scenarios probing malformed BatchSigners, signature-binding violations, stale-proof inner transactions, replay, over-size batches, and structural/serialization smuggling&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cross-Feature&lt;/td&gt;
&lt;td&gt;Interactions between Batch and Permission Delegation (XLS-75), Sponsored Fees &amp;amp; Reserves, Confidential MPT (XLS-96), and Tickets&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;End-to-End&lt;/td&gt;
&lt;td&gt;Full flows spanning batch assembly, multi-party signing, submission, atomic commit/rollback, and downstream metadata validation via xrpld&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;
&lt;h2&gt;
  
  
  4. Test Results Summary
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Testing Type&lt;/th&gt;
&lt;th&gt;Total Tests&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Batch — Core Functional (execution modes, multi-account/multi-sign, tickets/replay/metadata, vault/loan/transaction types, signature &amp;amp; structural validation)&lt;/td&gt;
&lt;td&gt;106&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Batch — Adversarial / Security&lt;/td&gt;
&lt;td&gt;56&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Batch — Cross-Feature (Delegation, Sponsorship, Confidential MPT)&lt;/td&gt;
&lt;td&gt;27&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Batch — Total&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;189&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Regression — xrpld (full suite)&lt;/td&gt;
&lt;td&gt;5,088&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Testcases: &lt;a href="https://dev.to/ripplexdev/batch-transaction-testcases-1klf"&gt;https://dev.to/ripplexdev/batch-transaction-testcases-1klf&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Feature commit history:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;- Introduced as  Batch  in PR #5060 (2a61aee562, merged 2025-05-23)
- Disabled in v3.1.1 due to a critical bug discovered in the original implementation
- Replaced by  BatchV1_1  (hardened BatchSigners + signing preimage) in PR #6446 (86d8b244d6, merged 2026-07-01)
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Related xrpld changes covered in this round:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/5060" rel="noopener noreferrer"&gt;#5060&lt;/a&gt; — Initial Batch (XLS-56) feature — four execution modes, &lt;code&gt;parentBatchId&lt;/code&gt; propagation&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/6446" rel="noopener noreferrer"&gt;#6446&lt;/a&gt; — BatchV1_1 — supersedes &lt;code&gt;Batch&lt;/code&gt;/&lt;code&gt;fixBatchInnerSigs&lt;/code&gt;, tightens BatchSigner authorization and signing-message construction&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/7736" rel="noopener noreferrer"&gt;#7736&lt;/a&gt; — enforces &lt;code&gt;kMaxBatchTxCount&lt;/code&gt; (rejects &amp;gt;8 inner transactions at STTx construction)&lt;/li&gt;
&lt;li&gt;PR &lt;a href="https://github.com/XRPLF/rippled/pull/7279" rel="noopener noreferrer"&gt;#7279&lt;/a&gt; — delegated Confidential MPT inner-transaction cases inside Batch&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Related specification change:&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;XRPL-Standards updated XLS-0056 to align with the BatchV1_1 implementation (BatchSigners sorting/signing payload, &lt;code&gt;tfInnerBatchTxn&lt;/code&gt; common-field semantics)&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  5. Bugs Reported
&lt;/h2&gt;

&lt;p&gt;All internal bugs are fixed and there are no Critical Open bugs.&lt;/p&gt;

&lt;h2&gt;
  
  
  6. Conclusion
&lt;/h2&gt;

&lt;p&gt;Batch (XLS-56) Transactions have now been exercised across 189 dedicated tests spanning core functional coverage (four execution modes, multi-account/multi-sign authorization, tickets/replay/metadata, and inner transaction-type coverage), adversarial / security scenarios, and cross-feature interactions with Permission Delegation, Sponsored Fees &amp;amp; Reserves, and Confidential MPT. In addition, the full xrpld (5,088) regression suite has been executed to confirm no downstream breakage.&lt;/p&gt;

&lt;p&gt;All security-critical properties documented in XLS-56 have been validated, including:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Correct atomic commit/rollback semantics for every execution mode (&lt;code&gt;AllOrNothing&lt;/code&gt; rolls the whole batch back on any inner failure; &lt;code&gt;OnlyOne&lt;/code&gt;, &lt;code&gt;UntilFailure&lt;/code&gt;, and &lt;code&gt;Independent&lt;/code&gt; apply their defined subsets)&lt;/li&gt;
&lt;li&gt;Enforcement of the 2–8 inner-transaction bounds and rejection of over-size batches before preflight work (PR &lt;a href="https://github.com/XRPLF/rippled/pull/7736" rel="noopener noreferrer"&gt;#7736&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;Correct BatchSigner authorization under the hardened &lt;code&gt;BatchV1_1&lt;/code&gt; signing preimage, including canonical ascending signer ordering and multi-signed BatchSigners&lt;/li&gt;
&lt;li&gt;Correct interaction with delegated authority, sponsorship, tickets, and Confidential MPT inner transactions, including stale-proof rollback under &lt;code&gt;AllOrNothing&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Accurate inner-transaction metadata (&lt;code&gt;sfParentBatchID&lt;/code&gt;) and same-ledger inclusion reporting&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The feature is considered ready for production use at the tested commit level. Future updates to this report will follow any material additions to the Batch surface.&lt;/p&gt;

</description>
      <category>devops</category>
      <category>softwaredevelopment</category>
      <category>testing</category>
    </item>
    <item>
      <title>Batch Transaction - Testcases</title>
      <dc:creator>Ramkumar SG</dc:creator>
      <pubDate>Tue, 08 Sep 2026 21:40:47 +0000</pubDate>
      <link>https://dev.to/ripplexdev/batch-transaction-testcases-1klf</link>
      <guid>https://dev.to/ripplexdev/batch-transaction-testcases-1klf</guid>
      <description>&lt;p&gt;Automated test coverage for the &lt;strong&gt;Batch&lt;/strong&gt; transaction feature (XLS-56), grouped by the invariant or execution mode each test exercises with &lt;strong&gt;189 tests.&lt;/strong&gt;&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Category&lt;/th&gt;
&lt;th&gt;Test Count&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Execution Modes&lt;/td&gt;
&lt;td&gt;42&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Multi-Account &amp;amp; Multi-Sign&lt;/td&gt;
&lt;td&gt;23&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Tickets, Replay &amp;amp; Metadata&lt;/td&gt;
&lt;td&gt;7&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Vault, Loan &amp;amp; Transaction Types&lt;/td&gt;
&lt;td&gt;24&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Signature &amp;amp; Structural Validation&lt;/td&gt;
&lt;td&gt;10&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Security &amp;amp; Adversarial&lt;/td&gt;
&lt;td&gt;56&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cross-Feature Interactions&lt;/td&gt;
&lt;td&gt;27&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Total&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;189&lt;/strong&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  1. Execution Modes
&lt;/h2&gt;

&lt;h3&gt;
  
  
  AllOrNothing
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch allornothing all payments succeed&lt;/li&gt;
&lt;li&gt;Test Batch allornothing submit batch multiple times&lt;/li&gt;
&lt;li&gt;Test Batch allornothing one payment fails&lt;/li&gt;
&lt;li&gt;Test Batch allornothing all payments fail&lt;/li&gt;
&lt;li&gt;Test Batch allornothing mixed transaction types&lt;/li&gt;
&lt;li&gt;Test Batch allornothing fee calculation&lt;/li&gt;
&lt;li&gt;Test Batch allornothing max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch allornothing more than max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch allornothing cash same check multiple times&lt;/li&gt;
&lt;li&gt;Test Batch allornothing fail and then succeed&lt;/li&gt;
&lt;li&gt;Test Batch allornothing with tickets&lt;/li&gt;
&lt;li&gt;Test Batch allornothing deep rollback on late inner failure&lt;/li&gt;
&lt;li&gt;Test Batch allornothing value conservation&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  OnlyOne
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch onlyone first succeeds&lt;/li&gt;
&lt;li&gt;Test Batch onlyone first fails second succeeds&lt;/li&gt;
&lt;li&gt;Test Batch onlyone all fail&lt;/li&gt;
&lt;li&gt;Test Batch onlyone offer priority&lt;/li&gt;
&lt;li&gt;Test Batch onlyone max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch onlyone more than max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch onlyone cash same check multiple times&lt;/li&gt;
&lt;li&gt;Test Batch onlyone fail and then succeed&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  UntilFailure
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch untilfailure all succeed&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure stops at first&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure stops at second&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure stops at third&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure sequential setup&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure progressive payments&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure mixed success failure pattern&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure more than max inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure cash same check multiple times&lt;/li&gt;
&lt;li&gt;Test Batch untilfailure fail and then succeed&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Independent
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch independent all succeed&lt;/li&gt;
&lt;li&gt;Test Batch independent some fail&lt;/li&gt;
&lt;li&gt;Test Batch independent all fail&lt;/li&gt;
&lt;li&gt;Test Batch independent parallel operations&lt;/li&gt;
&lt;li&gt;Test Batch independent mixed transaction types&lt;/li&gt;
&lt;li&gt;Test Batch independent account setup operations&lt;/li&gt;
&lt;li&gt;Test Batch independent max transactions&lt;/li&gt;
&lt;li&gt;Test Batch independent more than max transactions&lt;/li&gt;
&lt;li&gt;Test Batch independent cash same check multiple times&lt;/li&gt;
&lt;li&gt;Test Batch independent fail and then succeed&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  2. Multi-Account &amp;amp; Multi-Sign
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Multi-Account
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch multiaccount workflow all payments succeed&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount atomic swap xrp for xrp&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount atomic swap xrp for token&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount circular three way swap&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount with submitter not in inner txns&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount without signers&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount with max signers&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount with more than max signers&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount txn1 fails txn2 succeeds&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount create check succeeds cash check fails&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount delete accounts too soon with other account as signer&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount delete accounts after flag ledger with other account as signer&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount delete submitting account after flag ledger with no signer&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount delete submitting account after flag ledger with signer&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount alice creates offer bob creates offer&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount create nftokenmint nftokenoffercreate fails reverts nftokenmint&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount exploit steal funds&lt;/li&gt;
&lt;li&gt;Test Batch multiaccount ticket based outer succeeds&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Multi-Sign
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test multisigned outer batch succeeds&lt;/li&gt;
&lt;li&gt;Test multisigned outer with multisigned batch signer succeeds&lt;/li&gt;
&lt;li&gt;Test multisigned outer below quorum rejected tefBAD_QUORUM&lt;/li&gt;
&lt;li&gt;Test multisigned batch signer below quorum rejected tefBAD_QUORUM&lt;/li&gt;
&lt;li&gt;Test multisigned outer with unauthorized signer rejected tefBAD_SIGNATURE&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  3. Tickets, Replay &amp;amp; Metadata
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Tickets
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch inner txns consume tickets succeeds&lt;/li&gt;
&lt;li&gt;Test Batch inner tickets used out of order succeeds&lt;/li&gt;
&lt;li&gt;Test Batch inner with sequence and ticket rejected temSEQ_AND_TICKET&lt;/li&gt;
&lt;li&gt;Test Batch duplicate ticket same account rejected temREDUNDANT&lt;/li&gt;
&lt;li&gt;Test Batch nonexistent ticket rolls back allornothing&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Replay
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch resubmission is rejected and pays inner once&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Metadata
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch inner txns carry parent batch id and share ledger&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  4. Vault, Loan &amp;amp; Transaction Types
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Single-Asset Vault (SAV)
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch with vault create&lt;/li&gt;
&lt;li&gt;Test Batch with vault set&lt;/li&gt;
&lt;li&gt;Test Batch with vault deposit&lt;/li&gt;
&lt;li&gt;Test Batch with vault withdraw&lt;/li&gt;
&lt;li&gt;Test Batch with vault delete&lt;/li&gt;
&lt;li&gt;Test Batch with vault clawback&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  LoanBroker / Loan
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch with loan broker set&lt;/li&gt;
&lt;li&gt;Test Batch with loan broker delete&lt;/li&gt;
&lt;li&gt;Test Batch with loan broker cover deposit&lt;/li&gt;
&lt;li&gt;Test Batch with loan broker cover withdraw&lt;/li&gt;
&lt;li&gt;Test Batch with loan broker cover clawback&lt;/li&gt;
&lt;li&gt;Test Batch with loan set&lt;/li&gt;
&lt;li&gt;Test Batch with loan pay&lt;/li&gt;
&lt;li&gt;Test Batch with loan delete&lt;/li&gt;
&lt;li&gt;Test Batch with loan manage&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Supported Transaction Types
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch transaction type trustset&lt;/li&gt;
&lt;li&gt;Test Batch transaction type offercreate&lt;/li&gt;
&lt;li&gt;Test Batch transaction type account set&lt;/li&gt;
&lt;li&gt;Test Batch transaction type escrow create&lt;/li&gt;
&lt;li&gt;Test Batch transaction type check create&lt;/li&gt;
&lt;li&gt;Test Batch transaction type nftokenmint&lt;/li&gt;
&lt;li&gt;Test Batch with ticket create transaction&lt;/li&gt;
&lt;li&gt;Test Batch transaction type signerlistset&lt;/li&gt;
&lt;li&gt;Test Batch transaction type depositpreauth&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  5. Signature &amp;amp; Structural Validation (V1.1 Readiness)
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Readiness / Preimage
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch v1.1 preimage matches pr1010 single&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 preimage matches pr1010 multi&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 preimage layout&lt;/li&gt;
&lt;li&gt;Test batchsigners signature binds outer account sequence and signer account&lt;/li&gt;
&lt;li&gt;Test Batch signers sorted by account id&lt;/li&gt;
&lt;li&gt;Test sdk encode for signing batch natively binds account and sequence&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Preflight Error Ordering
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch inner tx with negative fee rejected at preflight&lt;/li&gt;
&lt;li&gt;Test Batch inner tx with positive xrp fee rejected at preflight&lt;/li&gt;
&lt;li&gt;Test Batch inner tx with iou fee rejected at preflight&lt;/li&gt;
&lt;li&gt;Test Batch inner tx with zero fee passes fee preflight&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  6. Security &amp;amp; Adversarial
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Structural / Flag Enforcement
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch empty raw transactions&lt;/li&gt;
&lt;li&gt;Test Batch with one raw transaction&lt;/li&gt;
&lt;li&gt;Test Batch with all inner transactions having no tfInnerBatchTxn flag&lt;/li&gt;
&lt;li&gt;Test Batch with some inner transactions having no tfInnerBatchTxn flag&lt;/li&gt;
&lt;li&gt;Test Batch with all inner transactions having incorrect tfInnerBatchTxn flag&lt;/li&gt;
&lt;li&gt;Test Batch with tfInnerBatchTxn flag set for outer transaction&lt;/li&gt;
&lt;li&gt;Test Batch with no mode set for outer transaction&lt;/li&gt;
&lt;li&gt;Test Batch with invalid mode set for outer transaction&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions having no fee field&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions having invalid fee values&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions having bad fee values&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with SigningPubKey&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with fee and SigningPubKey&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with TxnSignature&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with Signers&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with SponsorSignature&lt;/li&gt;
&lt;li&gt;Test Batch with inner transactions with CounterpartySignature&lt;/li&gt;
&lt;li&gt;Test Batch with invalid inner transactions&lt;/li&gt;
&lt;li&gt;Test Batch inside non empty batch&lt;/li&gt;
&lt;li&gt;Test Batch inside empty batch&lt;/li&gt;
&lt;li&gt;Test Batch inner txns with same sequence leaves no state change&lt;/li&gt;
&lt;li&gt;Test Batch inner txns with same sequence&lt;/li&gt;
&lt;li&gt;Test Batch inner txns with past sequence&lt;/li&gt;
&lt;li&gt;Test Batch inner txns with future not incremental sequences&lt;/li&gt;
&lt;li&gt;Test Batch with account does not exist&lt;/li&gt;
&lt;li&gt;Test Batch with deposit auth on account&lt;/li&gt;
&lt;li&gt;Test Batch with master key disabled on account&lt;/li&gt;
&lt;li&gt;Test Batch all payments with destination account not funded&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Adversarial Attack Matrix
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch with single inner tx returns temARRAY_EMPTY&lt;/li&gt;
&lt;li&gt;Test inner tx with non zero fee returns temBAD_FEE&lt;/li&gt;
&lt;li&gt;Test inner tx with non empty signing pubkey returns temBAD_REGKEY&lt;/li&gt;
&lt;li&gt;Test inner tx missing inner batch flag returns temINVALID_FLAG&lt;/li&gt;
&lt;li&gt;Test standalone payment with tfInnerBatchTxn flag rejected&lt;/li&gt;
&lt;li&gt;Test Batch with outer account in BatchSigners rejected&lt;/li&gt;
&lt;li&gt;Test Batch with duplicate BatchSigners rejected&lt;/li&gt;
&lt;li&gt;Test Batch with BatchSigner for unrelated account rejected&lt;/li&gt;
&lt;li&gt;Test Batch BatchSigners signature for different batch rejected&lt;/li&gt;
&lt;li&gt;Test Batch with single inner transaction rejected temARRAY_EMPTY&lt;/li&gt;
&lt;li&gt;Test Batch with nested inner batch rejected&lt;/li&gt;
&lt;li&gt;Test Batch with duplicate inner sequences rejected temREDUNDANT&lt;/li&gt;
&lt;li&gt;Test Batch inner setregularkey succeeds when outer signed by master&lt;/li&gt;
&lt;li&gt;Test Batch innerbatch flag on standalone payment rejected&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 oversized inner array rejected as tem&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 far oversized inner array survives eager hashing&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 far oversized inner array at scale survives eager hashing&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 batchsigner sig bound to outer sequence&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 batchsigner sig bound to outer account&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 unsorted batchsigners rejected&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 standalone tx with inner batch flag rejected&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 standalone oversized path stays tel&lt;/li&gt;
&lt;li&gt;Test Batch v1.1 batchsigners over kmax rejected&lt;/li&gt;
&lt;li&gt;Test non batch tx with raw transactions rejected at construction&lt;/li&gt;
&lt;li&gt;Test non batch raw transactions rejection does not crash server&lt;/li&gt;
&lt;li&gt;Test Batch outer with sfDelegate rejected by notdelegable template&lt;/li&gt;
&lt;li&gt;Test Batch inner notdelegable type with sfDelegate inner rejected&lt;/li&gt;
&lt;li&gt;Test Batch inner delegated without permission grant returns terNO_DELEGATE_PERMISSION&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  7. Cross-Feature Interactions
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Delegation
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test delegate as outer account executes delegated inner succeeds&lt;/li&gt;
&lt;li&gt;Test delegate as batch signer authorizes delegated inner succeeds&lt;/li&gt;
&lt;li&gt;Test delegated inner without batch signer rejected temBAD_SIGNER&lt;/li&gt;
&lt;li&gt;Test delegated inner signed by account holder rejected temBAD_SIGNER&lt;/li&gt;
&lt;li&gt;Test delegate without permission inner fails and batch rolls back&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Sponsorship
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test outer batch with reserve sponsorship flag rejected temINVALID_FLAG&lt;/li&gt;
&lt;li&gt;Test inner batch txn with fee sponsorship rejected temINVALID_FLAG&lt;/li&gt;
&lt;li&gt;Test inner batch txn with sponsor signature material rejected temBAD_SIGNATURE&lt;/li&gt;
&lt;li&gt;Test prefunded fee sponsored outer batch succeeds&lt;/li&gt;
&lt;/ol&gt;

&lt;h3&gt;
  
  
  Confidential MPT
&lt;/h3&gt;

&lt;ol&gt;
&lt;li&gt;Test Batch confidential mpt two sends with stale proof allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt independent send with one failing&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt convert then merge inbox allornothing&lt;/li&gt;
&lt;li&gt;Test Batch delegated send missing permission allornothing&lt;/li&gt;
&lt;li&gt;Test Batch delegated send with delegate as outer needs no batch signer&lt;/li&gt;
&lt;li&gt;Test Batch delegated send with delegate as outer with batch signer&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt two chained sends same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt send then convert back same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt three chained sends same sender allornothing&lt;/li&gt;
&lt;li&gt;Test Batch confidential mpt chained send wrong predicted ciphertext fails&lt;/li&gt;
&lt;li&gt;Test Batch only one with two valid chained sends executes first only&lt;/li&gt;
&lt;li&gt;Test Batch only one first send to unconverted fails falls through to second&lt;/li&gt;
&lt;li&gt;Test Batch until failure first send commits then stale second fails&lt;/li&gt;
&lt;li&gt;Test Batch allornothing issuer lock then alice send rolls back both&lt;/li&gt;
&lt;li&gt;Test Batch independent three holders converts coa aggregate correct&lt;/li&gt;
&lt;li&gt;Test Batch allornothing inner revokes delegate then later inner uses it rolls back&lt;/li&gt;
&lt;li&gt;Test Batch two chained delegated sends from same delegator both commit&lt;/li&gt;
&lt;li&gt;Test delegated send after delegator loses domain membership fails&lt;/li&gt;
&lt;/ol&gt;

</description>
      <category>automation</category>
      <category>security</category>
      <category>software</category>
      <category>testing</category>
    </item>
  </channel>
</rss>
