<?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: Prilog</title>
    <description>The latest articles on DEV Community by Prilog (prilog).</description>
    <link>https://dev.to/prilog</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%2F15835%2Fb6f411d5-5ae0-4db4-8b01-5bfecabfedd8.png</url>
      <title>DEV Community: Prilog</title>
      <link>https://dev.to/prilog</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/prilog"/>
    <language>en</language>
    <item>
      <title>Guarding Vidstack's TimeSlider.Value After Client-Side Navigation</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:55:11 +0000</pubDate>
      <link>https://dev.to/prilog/guarding-vidstacks-timeslidervalue-after-client-side-navigation-4f70</link>
      <guid>https://dev.to/prilog/guarding-vidstacks-timeslidervalue-after-client-side-navigation-4f70</guid>
      <description>&lt;p&gt;A method called on a component that was already torn down.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;In a Next.js App Router production build, navigating client-side to a page with a Vidstack player could crash with &lt;code&gt;TypeError: e is not a function at SliderValue.getValueText&lt;/code&gt;. The method could run on a destroyed component instance, or before its slider state and format context were resolved. The fix returns an empty string in that case. It was merged as &lt;a href="https://github.com/vidstack/player/pull/1853" rel="noopener noreferrer"&gt;PR #1853&lt;/a&gt; and closed &lt;a href="https://github.com/vidstack/player/issues/1852" rel="noopener noreferrer"&gt;issue #1852&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;Issue #1852 included a minimal player and clear steps: a direct load or reload of &lt;code&gt;/player&lt;/code&gt; rendered correctly, but navigating there through &lt;code&gt;next/link&lt;/code&gt; triggered the route error boundary. The reporter confirmed it with &lt;code&gt;@vidstack/react@1.15.6&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;   &lt;span class="nd"&gt;@method&lt;/span&gt;
   &lt;span class="nf"&gt;getValueText&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="kc"&gt;this&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;scope&lt;/span&gt; &lt;span class="o"&gt;||&lt;/span&gt; &lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="kc"&gt;this&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;#&lt;span class="n"&gt;slider&lt;/span&gt; &lt;span class="o"&gt;||&lt;/span&gt; &lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="kc"&gt;this&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;#&lt;span class="n"&gt;format&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="s1"&gt;''&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The PR description notes this follows the existing &lt;code&gt;if (!this.scope) return;&lt;/code&gt; pattern in &lt;code&gt;slider-chapters.ts&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/vidstack/player" rel="noopener noreferrer"&gt;vidstack/player&lt;/a&gt;, 3,720 stars (Nov 3, 2026), TypeScript, MIT&lt;/li&gt;
&lt;li&gt;Files changed: 1&lt;/li&gt;
&lt;li&gt;Lines: +2 / -0&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened and merged: Aug 21, 2026. Time to merge: 58 minutes&lt;/li&gt;
&lt;li&gt;Review: merged by a maintainer&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Did the PR include the claimed regression test?
&lt;/h3&gt;

&lt;p&gt;The description claimed a regression test, but the merged diff contains only the two-line guard and no test.&lt;/p&gt;

&lt;h3&gt;
  
  
  Is Vidstack still maintained?
&lt;/h3&gt;

&lt;p&gt;As recorded in the supplied research on Nov 3, 2026, the repository described itself as "Deprecated in favour of Video.js 10; security fixes only until January 2028."&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/vidstack/player/pull/1853" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/vidstack/player/issues/1852" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Components that outlive their lifecycle should fail quietly, not take the page down with them.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/vidstack-timeslider-value-destroyed-instance-guard/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>typescript</category>
      <category>react</category>
      <category>nextjs</category>
      <category>debugging</category>
    </item>
    <item>
      <title>Connecting TinaCMS's Boolean Toggle to Its Label</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:55:04 +0000</pubDate>
      <link>https://dev.to/prilog/connecting-tinacmss-boolean-toggle-to-its-label-1p1m</link>
      <guid>https://dev.to/prilog/connecting-tinacmss-boolean-toggle-to-its-label-1p1m</guid>
      <description>&lt;p&gt;A prop that never arrived, so screen readers got an unlabelled checkbox.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;TinaCMS's &lt;code&gt;boolean&lt;/code&gt; field renders a hidden checkbox behind a styled switch. Both the input's &lt;code&gt;id&lt;/code&gt; and the label's &lt;code&gt;htmlFor&lt;/code&gt; came from a &lt;code&gt;name&lt;/code&gt; prop that the field wrapper never passed, so both rendered empty and the label was not associated with any control. The fix falls back to &lt;code&gt;input.name&lt;/code&gt; when no &lt;code&gt;name&lt;/code&gt; prop is given. It was merged as &lt;a href="https://github.com/tinacms/tinacms/pull/7478" rel="noopener noreferrer"&gt;PR #7478&lt;/a&gt; and closed &lt;a href="https://github.com/tinacms/tinacms/issues/7473" rel="noopener noreferrer"&gt;issue #7473&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/tinacms/tinacms" rel="noopener noreferrer"&gt;TinaCMS&lt;/a&gt; is an open-source headless CMS with visual editing. Issue #7473 spelled out the effect: "Screen readers get an unlabelled checkbox, and the label is inert." Clicking the switch still worked, because the hidden input sits on top of it, which is why the bug went unnoticed visually.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight tsx"&gt;&lt;code&gt;&lt;span class="o"&gt;+&lt;/span&gt;  &lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="n"&gt;inputId&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;name&lt;/span&gt; &lt;span class="o"&gt;??&lt;/span&gt; &lt;span class="n"&gt;input&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;name&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;        &lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;ToggleInput&lt;/span&gt; &lt;span class="n"&gt;id&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;name&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="kd"&gt;type&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s1"&gt;'checkbox'&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;input&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="o"&gt;/&amp;gt;&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;        &lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="nc"&gt;ToggleInput&lt;/span&gt; &lt;span class="kd"&gt;type&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s1"&gt;'checkbox'&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;input&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="n"&gt;id&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;inputId&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="o"&gt;/&amp;gt;&lt;/span&gt;
&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;          &lt;span class="n"&gt;htmlFor&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;name&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;          &lt;span class="n"&gt;htmlFor&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;inputId&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The &lt;code&gt;name&lt;/code&gt; prop became optional, and &lt;code&gt;ToggleFieldDefinition&lt;/code&gt; is now exported so the types build can name it. A patch changeset was added for the &lt;code&gt;tinacms&lt;/code&gt; package.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/tinacms/tinacms" rel="noopener noreferrer"&gt;tinacms/tinacms&lt;/a&gt;, 13,864 stars (Nov 3, 2026), TypeScript, Apache-2.0&lt;/li&gt;
&lt;li&gt;Files changed: 2&lt;/li&gt;
&lt;li&gt;Lines: +10 / -4&lt;/li&gt;
&lt;li&gt;Commits: 3 (2 from the PR author, 1 maintainer merge of &lt;code&gt;main&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Sep 16, 2026. Time to merge: 25 days, 11 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer; CLA signed&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;The maintainer asked: "Could you please fix the CI and add a changeset?" The follow-up commit exported &lt;code&gt;ToggleFieldDefinition&lt;/code&gt;, which the types build had failed to name once &lt;code&gt;name&lt;/code&gt; became optional, and added the changeset. The maintainer approved with "LGTM" and merged.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Did the PR include the claimed unit test?
&lt;/h3&gt;

&lt;p&gt;The PR description claimed a unit test, but the merged diff contains the component change and a changeset, with no test file. The reviewing maintainer had already described the root cause in the issue.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/tinacms/tinacms/pull/7478" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/tinacms/tinacms/issues/7473" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;A control that works with a mouse but not with a screen reader is half-built. Wiring the label to its input fixes both.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/tinacms-boolean-toggle-label-id-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>typescript</category>
      <category>react</category>
      <category>a11y</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Stopping Kratix From Endlessly Requeuing Failed Delete Pipelines</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:58 +0000</pubDate>
      <link>https://dev.to/prilog/stopping-kratix-from-endlessly-requeuing-failed-delete-pipelines-21jp</link>
      <guid>https://dev.to/prilog/stopping-kratix-from-endlessly-requeuing-failed-delete-pipelines-21jp</guid>
      <description>&lt;p&gt;An error return that turned a terminal failure into a busy loop.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;When a delete pipeline for a Kratix Promise or resource request failed, the controller returned an error on every reconcile. Controller-runtime requeues on error, so the object was reconciled over and over without the pipeline ever rerunning. The fix treats a failed delete pipeline as settled: it still records the failure condition and event, logs it, and returns without an error. Recovery happens through a manual reconciliation, a pipeline change, or the existing &lt;code&gt;retryAfter&lt;/code&gt; mechanism. It was merged as &lt;a href="https://github.com/syntasso/kratix/pull/968" rel="noopener noreferrer"&gt;PR #968&lt;/a&gt; and closed &lt;a href="https://github.com/syntasso/kratix/issues/102" rel="noopener noreferrer"&gt;issue #102&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/syntasso/kratix" rel="noopener noreferrer"&gt;Kratix&lt;/a&gt; is an open-source framework for building internal platforms on Kubernetes. Issue #102, opened in April 2024, flagged the loop and its cost: "if many pipelines fail at once on the same cluster: the continual requeuing may end up using a chunk of the resources." The issue was framed as an open question: "It's unclear what the right behaviour is here (since returning an error will also requeue), so this ticket is a placeholder/RFC!"&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;

&lt;p&gt;The same pattern in both the Promise and dynamic resource-request controllers:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Is&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;err&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;workflow&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;ErrDeletePipelineFailed&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="n"&gt;logging&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Info&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;o&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;logger&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"resource delete workflow failed; waiting for a manual reconciliation"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;ctrl&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Result&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kc"&gt;nil&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Existing tests that expected the error were updated, and new cases assert that a further reconcile neither errors nor requeues.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/syntasso/kratix" rel="noopener noreferrer"&gt;syntasso/kratix&lt;/a&gt;, 785 stars (Nov 3, 2026), Go, Apache-2.0&lt;/li&gt;
&lt;li&gt;Files changed: 4&lt;/li&gt;
&lt;li&gt;Lines: +24 / -5&lt;/li&gt;
&lt;li&gt;Commits: 4 (1 from the PR author, 3 maintainer merges of &lt;code&gt;main&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Opened: Oct 10, 2026. Merged: Oct 20, 2026. Time to merge: 10 days, 1 hour&lt;/li&gt;
&lt;li&gt;Review: merged by a maintainer after the CLA was signed&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;The maintainer wrote "Thanks for your contribution @Roshan931 ! If you could sign the CLA, we can get this merged!" Once it was signed, they merged it with "Thank you!"&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  How do I retry a failed delete pipeline now?
&lt;/h3&gt;

&lt;p&gt;Per the PR description, through a manual reconciliation or a pipeline change, which restarts the workflow, or through the existing &lt;code&gt;retryAfter&lt;/code&gt; workflow control.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/syntasso/kratix/pull/968" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/syntasso/kratix/issues/102" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;In a controller, returning an error means "try again". When trying again can't help, record the failure and wait for a signal that something changed.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/kratix-failed-delete-pipeline-requeue-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>go</category>
      <category>kubernetes</category>
      <category>devops</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Fixing Empty Teaser Titles in Sulu 3 for Content Migrated From Sulu 2</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:51 +0000</pubDate>
      <link>https://dev.to/prilog/fixing-empty-teaser-titles-in-sulu-3-for-content-migrated-from-sulu-2-577h</link>
      <guid>https://dev.to/prilog/fixing-empty-teaser-titles-in-sulu-3-for-content-migrated-from-sulu-2-577h</guid>
      <description>&lt;p&gt;The difference between &lt;code&gt;null&lt;/code&gt; and &lt;code&gt;''&lt;/code&gt;, in a CMS teaser.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;Sulu's page and article teaser providers fell back to the content title only when the excerpt title was missing (&lt;code&gt;null&lt;/code&gt;). Content migrated from Sulu 2 stores an unset excerpt title as an empty string, so those teasers ended up with no title. It showed most on pages turned into internal links, whose teaser uses the link target's excerpt. The fix treats an empty string the same as a missing value. It was merged as &lt;a href="https://github.com/sulu/sulu/pull/9039" rel="noopener noreferrer"&gt;PR #9039&lt;/a&gt; and closed &lt;a href="https://github.com/sulu/sulu/issues/8766" rel="noopener noreferrer"&gt;issue #8766&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/sulu/sulu" rel="noopener noreferrer"&gt;Sulu&lt;/a&gt; is a Symfony-based content management system. Issue #8766, on Sulu 3.0.x, reported that a &lt;code&gt;teaser_selection&lt;/code&gt; returned empty data for pages later changed into internal links, and pointed at &lt;code&gt;PageTeaserProvider::resolveTitle&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Root cause
&lt;/h2&gt;

&lt;p&gt;The &lt;code&gt;??&lt;/code&gt; operator only catches &lt;code&gt;null&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight php"&gt;&lt;code&gt;&lt;span class="nv"&gt;$title&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nv"&gt;$dimensionContent&lt;/span&gt;&lt;span class="o"&gt;-&amp;gt;&lt;/span&gt;&lt;span class="nf"&gt;getExcerptTitle&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;??&lt;/span&gt; &lt;span class="nv"&gt;$dimensionContent&lt;/span&gt;&lt;span class="o"&gt;-&amp;gt;&lt;/span&gt;&lt;span class="nf"&gt;getTitle&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;An empty string passed straight through, and the method then returned &lt;code&gt;null&lt;/code&gt; for the empty title. As the PR description notes, both providers already treated an empty description and an empty "more" text this way; the title was the odd one out.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;

&lt;p&gt;The same change in &lt;code&gt;PageTeaserProvider&lt;/code&gt; and &lt;code&gt;ArticleTeaserProvider&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight php"&gt;&lt;code&gt;&lt;span class="nv"&gt;$excerptTitle&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nv"&gt;$dimensionContent&lt;/span&gt;&lt;span class="o"&gt;-&amp;gt;&lt;/span&gt;&lt;span class="nf"&gt;getExcerptTitle&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="nv"&gt;$title&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;''&lt;/span&gt; &lt;span class="o"&gt;!==&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nv"&gt;$excerptTitle&lt;/span&gt; &lt;span class="o"&gt;??&lt;/span&gt; &lt;span class="s1"&gt;''&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;?&lt;/span&gt; &lt;span class="nv"&gt;$excerptTitle&lt;/span&gt; &lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nv"&gt;$dimensionContent&lt;/span&gt;&lt;span class="o"&gt;-&amp;gt;&lt;/span&gt;&lt;span class="nf"&gt;getTitle&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Three functional tests cover it: an empty excerpt title on a page, on an article, and an internal link to a page whose excerpt title is empty.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/sulu/sulu" rel="noopener noreferrer"&gt;sulu/sulu&lt;/a&gt;, 1,363 stars (Nov 3, 2026), PHP, MIT&lt;/li&gt;
&lt;li&gt;Files changed: 4&lt;/li&gt;
&lt;li&gt;Lines: +87 / -2&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Sep 22, 2026. Time to merge: 31 days, 14 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;A maintainer approved and merged it, replying "thank you! :)".&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Do I need to re-save migrated content?
&lt;/h3&gt;

&lt;p&gt;No. The fallback happens when the teaser is resolved.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/sulu/sulu/pull/9039" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/sulu/sulu/issues/8766" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Migrations leave data in shapes the new code didn't expect. Treating "empty" and "missing" the same where they mean the same keeps migrated sites looking right.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/sulu-empty-excerpt-teaser-title-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>php</category>
      <category>symfony</category>
      <category>cms</category>
      <category>testing</category>
    </item>
    <item>
      <title>When % Isn't a Format Verb: Fixing LIKE Patterns in StackQL Views</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:45 +0000</pubDate>
      <link>https://dev.to/prilog/when-isnt-a-format-verb-fixing-like-patterns-in-stackql-views-ic5</link>
      <guid>https://dev.to/prilog/when-isnt-a-format-verb-fixing-like-patterns-in-stackql-views-ic5</guid>
      <description>&lt;p&gt;User SQL passed through &lt;code&gt;fmt.Sprintf&lt;/code&gt;, where every &lt;code&gt;%&lt;/code&gt; means something.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;StackQL builds the final SQL for views and subqueries by substituting the inner query into a template with &lt;code&gt;fmt.Sprintf&lt;/code&gt;. A &lt;code&gt;LIKE '%TLS12%'&lt;/code&gt; pattern in a query on a view therefore contained characters that Go read as format verbs, and the result failed with &lt;code&gt;sql packet preparation error: near "(": syntax error&lt;/code&gt;. The fix replaces the printf-based substitution with explicit placeholder tokens and a literal-safe expander. It was merged as &lt;a href="https://github.com/stackql/stackql/pull/723" rel="noopener noreferrer"&gt;PR #723&lt;/a&gt; and closed &lt;a href="https://github.com/stackql/stackql/issues/472" rel="noopener noreferrer"&gt;issue #472&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;Issue #472, opened in October 2024, showed a query on the &lt;code&gt;azure.network.vw_application_gateways&lt;/code&gt; view with &lt;code&gt;ssl_policy NOT LIKE '%TLS12%'&lt;/code&gt; failing, while the equivalent query on the base resource worked.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;

&lt;p&gt;Templates now carry named tokens instead of &lt;code&gt;%s&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="nc"&gt;IndirectQueryPlaceholder&lt;/span&gt;   &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"__iql__placeholder_indirect_query__"&lt;/span&gt;
&lt;span class="nc"&gt;ControlOnClausePlaceholder&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"__iql__placeholder_control_on_clause__"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="o"&gt;-&lt;/span&gt;       &lt;span class="n"&gt;query&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;fmt&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Sprintf&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;ca&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;GetQuery&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;childQueryStrings&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;       &lt;span class="n"&gt;query&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;textutil&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;ExpandPlaceholders&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;query&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;textutil&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;IndirectQueryPlaceholder&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;childQueryStrings&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;ExpandPlaceholders&lt;/code&gt; replaces tokens left to right and never interprets the replacement text. Unit tests include a template with &lt;code&gt;NOT LIKE '%TLS12%'&lt;/code&gt;, and a Robot Framework scenario was updated.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/stackql/stackql" rel="noopener noreferrer"&gt;stackql/stackql&lt;/a&gt;, 1,074 stars (Nov 3, 2026), Go, MIT&lt;/li&gt;
&lt;li&gt;Files changed: 8&lt;/li&gt;
&lt;li&gt;Lines: +164 / -30&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Aug 22, 2026. Time to merge: 16 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;The maintainer approved and merged it alongside &lt;a href="https://github.com/stackql/stackql/pull/722" rel="noopener noreferrer"&gt;PR #722&lt;/a&gt;, with "thanks again".&lt;/p&gt;

&lt;h2&gt;
  
  
  Takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Never use user text as a format string.&lt;/strong&gt; Any &lt;code&gt;%&lt;/code&gt; in a SQL literal is fair game for &lt;code&gt;Sprintf&lt;/code&gt; once it lands in the template.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Use distinctive placeholder tokens.&lt;/strong&gt; A long, distinctive placeholder reduces accidental collisions with ordinary SQL text.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/stackql/stackql/pull/723" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/stackql/stackql/issues/472" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Templating and user input should never share a syntax. Separating them makes the rewriter safe for whatever SQL people write.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/stackql-view-like-pattern-sprintf-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>go</category>
      <category>sql</category>
      <category>debugging</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Letting StackQL Project Functions Over Subqueries: Fixing JSON_EXTRACT Rewrites</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:38 +0000</pubDate>
      <link>https://dev.to/prilog/letting-stackql-project-functions-over-subqueries-fixing-jsonextract-rewrites-o51</link>
      <guid>https://dev.to/prilog/letting-stackql-project-functions-over-subqueries-fixing-jsonextract-rewrites-o51</guid>
      <description>&lt;p&gt;A rewriter that tried to resolve every projection as a column, even when it was a function.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;Selecting an expression such as &lt;code&gt;JSON_EXTRACT(contact_method, '$.html_url')&lt;/code&gt; from a subquery failed in StackQL with "query rewriting for indirection: cannot find col". The rewriter only let through projections that looked like &lt;code&gt;json_each&lt;/code&gt; values or &lt;code&gt;OR&lt;/code&gt; expressions; everything else had to match a subquery column by name. The fix treats any projection that isn't a bare column reference as an expression and passes it to the SQL backend. Bare column names that don't exist still error as before. It was merged as &lt;a href="https://github.com/stackql/stackql/pull/722" rel="noopener noreferrer"&gt;PR #722&lt;/a&gt; and closed &lt;a href="https://github.com/stackql/stackql/issues/352" rel="noopener noreferrer"&gt;issue #352&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/stackql/stackql" rel="noopener noreferrer"&gt;StackQL&lt;/a&gt; lets you query cloud and SaaS resources with SQL. Issue #352, opened in December 2023, showed a PagerDuty query that unpacked contact methods with &lt;code&gt;json_each&lt;/code&gt; in a subquery and then used &lt;code&gt;JSON_EXTRACT&lt;/code&gt; in the outer select. The rewriter couldn't find a column with the expression's inferred name and aborted.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="k"&gt;func&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;v&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;standardQueryRewriteAstVisitor&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="nf"&gt;isAnonymous&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;col&lt;/span&gt; &lt;span class="n"&gt;parserutil&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;ColumnHandle&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="kt"&gt;bool&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;   &lt;span class="k"&gt;switch&lt;/span&gt; &lt;span class="n"&gt;col&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Expr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="k"&gt;type&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;   &lt;span class="k"&gt;case&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;sqlparser&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;OrExpr&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;       &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;   &lt;span class="k"&gt;default&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;       &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;false&lt;/span&gt;
&lt;span class="o"&gt;-&lt;/span&gt;   &lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;&lt;span class="k"&gt;func&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;v&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;standardQueryRewriteAstVisitor&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="nf"&gt;isExpressionProjection&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;node&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;sqlparser&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;AliasedExpr&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="kt"&gt;bool&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;   &lt;span class="n"&gt;_&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;isColumnReference&lt;/span&gt; &lt;span class="o"&gt;:=&lt;/span&gt; &lt;span class="n"&gt;node&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Expr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;sqlparser&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;ColName&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;   &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="n"&gt;isColumnReference&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Two Robot Framework scenarios cover it: a &lt;code&gt;json_extract&lt;/code&gt; over a subquery (on the SQLite backend), and a &lt;code&gt;coalesce&lt;/code&gt; with a literal leading argument (on both backends).&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/stackql/stackql" rel="noopener noreferrer"&gt;stackql/stackql&lt;/a&gt;, 1,074 stars (Nov 3, 2026), Go, MIT&lt;/li&gt;
&lt;li&gt;Files changed: 2&lt;/li&gt;
&lt;li&gt;Lines: +39 / -9&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Aug 22, 2026. Time to merge: 16 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;A maintainer approved and merged it with "thank you for your contribution".&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/stackql/stackql/pull/722" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/stackql/stackql/issues/352" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;A rewriter should only resolve what it needs to. Treating expressions as opaque lets the database do the work it already knows how to do.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/stackql-subquery-function-projection-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>go</category>
      <category>sql</category>
      <category>testing</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Unfreezing Show My Cursor in RustDesk's View Only Mode</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:32 +0000</pubDate>
      <link>https://dev.to/prilog/unfreezing-show-my-cursor-in-rustdesks-view-only-mode-1k81</link>
      <guid>https://dev.to/prilog/unfreezing-show-my-cursor-in-rustdesks-view-only-mode-1k81</guid>
      <description>&lt;p&gt;A guard meant for controlling sessions, applied to a session that never controls.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;With View Only and Show My Cursor enabled, RustDesk forwards the observer's pointer so the remote side can see it. But those events still went through the peer-control protection check. When the remote user moved their mouse, the check treated the observer as contending for control and stopped sending updates, so the observer cursor froze. The fix skips that check in View Only plus Show My Cursor sessions, and stops local pointer moves from taking over the client-side cursor display in View Only mode. It was merged as &lt;a href="https://github.com/rustdesk/rustdesk/pull/15936" rel="noopener noreferrer"&gt;PR #15936&lt;/a&gt; and closed &lt;a href="https://github.com/rustdesk/rustdesk/issues/15214" rel="noopener noreferrer"&gt;issue #15214&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/rustdesk/rustdesk" rel="noopener noreferrer"&gt;RustDesk&lt;/a&gt; is an open-source, self-hostable remote desktop application. Issue #15214, on RustDesk 1.4.7 between two Windows 11 machines, described the observer cursor freezing "when the remote user starts moving their mouse", and only recovering after moving the local mouse out of the window and back in.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;

&lt;p&gt;The change is in the Flutter client:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight dart"&gt;&lt;code&gt;  &lt;span class="kt"&gt;bool&lt;/span&gt; &lt;span class="nf"&gt;_checkPeerControlProtected&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;double&lt;/span&gt; &lt;span class="n"&gt;x&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;double&lt;/span&gt; &lt;span class="n"&gt;y&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;isViewOnly&lt;/span&gt; &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; &lt;span class="n"&gt;showMyCursor&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
      &lt;span class="n"&gt;lastMousePos&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;ui&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Offset&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;x&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;y&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
      &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight dart"&gt;&lt;code&gt;  &lt;span class="kt"&gt;void&lt;/span&gt; &lt;span class="nf"&gt;updateLocalCursor&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;double&lt;/span&gt; &lt;span class="n"&gt;x&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;double&lt;/span&gt; &lt;span class="n"&gt;y&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;parent&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;target&lt;/span&gt;&lt;span class="o"&gt;?&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;ffiModel&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;viewOnly&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;return&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/rustdesk/rustdesk" rel="noopener noreferrer"&gt;rustdesk/rustdesk&lt;/a&gt;, 128,025 stars (Nov 3, 2026), Rust (change in the Dart/Flutter client), AGPL-3.0&lt;/li&gt;
&lt;li&gt;Files changed: 2&lt;/li&gt;
&lt;li&gt;Lines: +6 / -0&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Aug 24, 2026. Time to merge: 2 days, 16 hours&lt;/li&gt;
&lt;li&gt;Review: approved by a maintainer and merged&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;Automated reviewers from CodeRabbit and Greptile commented first. Greptile raised a P1 concern that the early return could also let pointer button events bypass control arbitration. The final diff still contains that early return, and the public review does not establish that the concern was resolved. A maintainer approved the PR, and it was merged without further changes. The PR description states that no regression test was added because the affected logic depends on native bindings.&lt;/p&gt;

&lt;p&gt;The Nov 3, 2026 review notes also record missing DCO sign-off despite the contribution documentation requiring it.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Does this affect sessions that control the remote machine?
&lt;/h3&gt;

&lt;p&gt;The arbitration bypass is gated on View Only plus Show My Cursor, while the local cursor-display guard applies to all View Only sessions. The unresolved button-event review concern means the merge alone does not establish that control behavior is safe in every affected path.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/rustdesk/rustdesk/pull/15936" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/rustdesk/rustdesk/issues/15214" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;A guard should know which situations it protects. Observer cursor updates need to work without weakening input restrictions. The unresolved review concern needs separate verification before treating this patch as evidence that both goals are met.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/rustdesk-view-only-show-my-cursor-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>rust</category>
      <category>dart</category>
      <category>flutter</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Returning Joined Columns Under Restricted Access in pREST</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:26 +0000</pubDate>
      <link>https://dev.to/prilog/returning-joined-columns-under-restricted-access-in-prest-18fp</link>
      <guid>https://dev.to/prilog/returning-joined-columns-under-restricted-access-in-prest-18fp</guid>
      <description>&lt;p&gt;A permission check that only looked at one side of the JOIN, with regression coverage for denied joined tables.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;With &lt;code&gt;restrict = true&lt;/code&gt;, pREST builds the SELECT list from the permitted fields of the requested table only. Any &lt;code&gt;_join&lt;/code&gt; target lost all its columns, even when the configuration granted read access to them. The fix resolves the join target, checks its own table and field permissions for the same user and operation, and adds only the permitted columns, qualified by table name. Joined tables without read permission still contribute nothing. It was merged as &lt;a href="https://github.com/prest/prest/pull/1032" rel="noopener noreferrer"&gt;PR #1032&lt;/a&gt; and closed &lt;a href="https://github.com/prest/prest/issues/364" rel="noopener noreferrer"&gt;issue #364&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/prest/prest" rel="noopener noreferrer"&gt;pREST&lt;/a&gt; serves instant REST APIs over PostgreSQL and MySQL. Issue #364, opened in January 2020, gave a clear reproduction: an &lt;code&gt;employee&lt;/code&gt; and a &lt;code&gt;department&lt;/code&gt; table, both granted &lt;code&gt;read&lt;/code&gt; with explicit fields, and a request:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;/test/public/department?_join=inner:employee:department.emp_id:$eq:employee.id
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The response contained only department columns. "However if access restricted is false it works as normal."&lt;/p&gt;

&lt;h2&gt;
  
  
  Root cause
&lt;/h2&gt;

&lt;p&gt;From the PR description: "restrict=true builds the SELECT list from the main table field permissions only, dropping joined table columns".&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;

&lt;p&gt;A new helper looks up permissions for the joined table:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="k"&gt;func&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;adapter&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;postgres&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="nf"&gt;joinFieldsPermissions&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt;&lt;span class="n"&gt;http&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Request&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;database&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;schema&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;op&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;userName&lt;/span&gt; &lt;span class="kt"&gt;string&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;fields&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="kt"&gt;string&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joined&lt;/span&gt; &lt;span class="kt"&gt;bool&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="n"&gt;joinSchema&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joinTable&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;ok&lt;/span&gt; &lt;span class="o"&gt;:=&lt;/span&gt; &lt;span class="nf"&gt;joinTableByRequest&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;schema&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="n"&gt;ok&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="n"&gt;joined&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="o"&gt;!&lt;/span&gt;&lt;span class="n"&gt;adapter&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;TablePermissions&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;database&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joinSchema&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joinTable&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;op&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;userName&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="n"&gt;allowedFields&lt;/span&gt; &lt;span class="o"&gt;:=&lt;/span&gt; &lt;span class="n"&gt;adapter&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;fieldsByPermission&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;database&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joinSchema&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;joinTable&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;op&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;userName&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="nf"&gt;containsAsterisk&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;allowedFields&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="kt"&gt;string&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;joinTable&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="s2"&gt;".*"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="nf"&gt;qualifyFields&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;joinTable&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;allowedFields&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;When a join is present, main-table columns are qualified too, so they can't become ambiguous. &lt;code&gt;_select&lt;/code&gt; accepts qualified fields from either side, and &lt;code&gt;sanitizeSelectField&lt;/code&gt; gained validated support for &lt;code&gt;table.*&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/prest/prest" rel="noopener noreferrer"&gt;prest/prest&lt;/a&gt;, 4,630 stars (Nov 3, 2026), Go, MIT&lt;/li&gt;
&lt;li&gt;Files changed: 4&lt;/li&gt;
&lt;li&gt;Lines: +404 / -6&lt;/li&gt;
&lt;li&gt;Commits: 4 (an initial fix and three "address review feedback" commits)&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Sep 16, 2026. Time to merge: 25 days, 7 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer; CLA signed&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;The maintainer asked for two things: signing the CLA, and "dont forget to include integration tests for your solution". Both were done. The PR added &lt;code&gt;integration/postgres/controllers/join_acl_test.go&lt;/code&gt;, which runs against the auth-enabled server with restricted access and covers an unreadable joined table, a readable one, and a wildcard field grant. CodeRabbit flagged a critical authorization gap in the initial patch: a base-table wildcard could bypass joined-table checks. Follow-up commit &lt;code&gt;7290488&lt;/code&gt; moved join authorization ahead of wildcard handling, and the final diff adds &lt;code&gt;TestSelectJoin_WildcardFieldsStayOnRequestedTable&lt;/code&gt; to cover denied joined-table columns, including an explicit &lt;code&gt;_select&lt;/code&gt;. Three follow-up commits addressed review feedback. The maintainer approved with "Thanks @Roshan931 for the contribution".&lt;/p&gt;

&lt;h2&gt;
  
  
  Takeaways
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Access control has to follow the query.&lt;/strong&gt; Once a request can touch two tables, the permission check has to look at both.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Test the negative case.&lt;/strong&gt; One integration test checks that columns from a table the user can't read never appear.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Can a join now expose columns a user isn't allowed to read?
&lt;/h3&gt;

&lt;p&gt;The fix checks the joined table's permissions for the same user and operation. If the user lacks read permission, the joined table contributes no columns, and an integration test asserts that.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/prest/prest/pull/1032" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/prest/prest/issues/364" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Restricted mode should restrict, not break. Applying each table's own grants keeps JOINs working without widening access.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/prest-restricted-access-join-columns-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>go</category>
      <category>postgres</category>
      <category>security</category>
      <category>testing</category>
    </item>
    <item>
      <title>Restoring Microsoft Teams Notifications in Prefect for New Power Automate URLs</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:19 +0000</pubDate>
      <link>https://dev.to/prilog/restoring-microsoft-teams-notifications-in-prefect-for-new-power-automate-urls-8g2</link>
      <guid>https://dev.to/prilog/restoring-microsoft-teams-notifications-in-prefect-for-new-power-automate-urls-8g2</guid>
      <description>&lt;p&gt;Sometimes the fix is upstream already, and the work is pinning the right version and proving it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;Microsoft's newer Power Automate workflow URLs weren't supported by the version of &lt;code&gt;apprise&lt;/code&gt; that Prefect allowed, so Teams notifications from Prefect automations stopped working for new workflows. Support landed in &lt;code&gt;apprise&lt;/code&gt; 1.13.0. The fix raises Prefect's minimum &lt;code&gt;apprise&lt;/code&gt; version to 1.13.0 in both the server and client packages, updates the lockfile, and adds tests for both new URL shapes. It was merged as &lt;a href="https://github.com/PrefectHQ/prefect/pull/22908" rel="noopener noreferrer"&gt;PR #22908&lt;/a&gt; and closed &lt;a href="https://github.com/PrefectHQ/prefect/issues/22846" rel="noopener noreferrer"&gt;issue #22846&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/PrefectHQ/prefect" rel="noopener noreferrer"&gt;Prefect&lt;/a&gt; is a Python workflow orchestration framework. Its &lt;code&gt;MicrosoftTeamsWebhook&lt;/code&gt; notification block relies on &lt;a href="https://github.com/caronc/apprise" rel="noopener noreferrer"&gt;apprise&lt;/a&gt; to parse and send webhooks.&lt;/p&gt;

&lt;p&gt;Issue #22846, on Prefect 3.7.2, reported that newly created Teams workflows use a new URL format that the installed &lt;code&gt;apprise&lt;/code&gt; didn't support. The reporter had traced it upstream and offered a workaround: install &lt;code&gt;apprise&lt;/code&gt; from a specific git commit inside the Prefect server container.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight diff"&gt;&lt;code&gt;&lt;span class="gd"&gt;-    "apprise&amp;gt;=1.1.0,&amp;lt;2.0.0",&lt;/span&gt;
&lt;span class="gi"&gt;+    "apprise&amp;gt;=1.13.0,&amp;lt;2.0.0",&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The same floor was applied to &lt;code&gt;client/pyproject.toml&lt;/code&gt;, and &lt;code&gt;uv.lock&lt;/code&gt; moved from 1.12.0 to 1.13.0. Two new tests cover the URL shapes the issue described:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="no"&gt;POWER_AUTOMATE_URL&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"https://prod-NO.LOCATION.logic.azure.com:443/powerautomate/automations/direct/workflows/WFID/triggers/manual/paths/invoke?..."&lt;/span&gt;
&lt;span class="no"&gt;POWER_AUTOMATE_URL_WITH_ROUTING_ID&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"https://prod-NO.LOCATION.logic.azure.com:443/powerautomate/automations/direct/cu/12/workflows/WFID/triggers/manual/paths/invoke?..."&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Each test checks the &lt;code&gt;workflow://&lt;/code&gt; URL that Prefect hands to apprise, including &lt;code&gt;pa=yes&lt;/code&gt; and, for the second shape, &lt;code&gt;route=12&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/PrefectHQ/prefect" rel="noopener noreferrer"&gt;PrefectHQ/prefect&lt;/a&gt;, 24,107 stars (Nov 3, 2026), Python, Apache-2.0&lt;/li&gt;
&lt;li&gt;Files changed: 4 (2 dependency manifests, 1 lockfile, 1 test file)&lt;/li&gt;
&lt;li&gt;Lines: +46 / -6&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Aug 26, 2026. Time to merge: 4 days, 23 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer; automated benchmarks reported no performance change&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;A maintainer approved and merged the PR without requesting changes. An automated benchmark check reported that the change would "not alter performance."&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Is this a code change in Prefect's notification logic?
&lt;/h3&gt;

&lt;p&gt;No. The fix is a dependency floor plus regression tests. The URL parsing itself lives in apprise 1.13.0.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/PrefectHQ/prefect/pull/22908" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/PrefectHQ/prefect/issues/22846" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Users had a workaround that patched a running container. A version floor and two tests replace it with something that survives a rebuild.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/prefect-teams-power-automate-webhook-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>python</category>
      <category>devops</category>
      <category>testing</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Fixing a Test That Could Never Pass: Portkey's Terraform Provider</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:13 +0000</pubDate>
      <link>https://dev.to/prilog/fixing-a-test-that-could-never-pass-portkeys-terraform-provider-2d73</link>
      <guid>https://dev.to/prilog/fixing-a-test-that-could-never-pass-portkeys-terraform-provider-2d73</guid>
      <description>&lt;p&gt;Not every bug is in production code. This one was in an acceptance test.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;An acceptance test in Portkey's Terraform provider asserted that a SCIM mapping's &lt;code&gt;workspace_id&lt;/code&gt;, which the API returns as a UUID, equals the workspace resource's &lt;code&gt;id&lt;/code&gt;, which is a slug. That check could never pass. The fix replaces it with a UUID-shape assertion. It was merged as &lt;a href="https://github.com/Portkey-AI/terraform-provider-portkey/pull/84" rel="noopener noreferrer"&gt;PR #84&lt;/a&gt; and closed &lt;a href="https://github.com/Portkey-AI/terraform-provider-portkey/issues/63" rel="noopener noreferrer"&gt;issue #63&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;Issue #63, filed by a maintainer, laid it out. &lt;code&gt;TestAccScimWorkspaceMappingsDataSource_withCreatedMapping&lt;/code&gt; failed against the live API:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Attribute 'mappings.0.workspace_id' expected "ws-tf-acc-329a4f", got "d4f92461-0e9b-4c91-9b70-eb2d19c099ed"
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The issue called it "a test bug, not a provider bug." The data source correctly reports the API's UUID, while the resource keeps the user's slug. It also noted that CI doesn't run acceptance tests, which is how the check went unnoticed.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight diff"&gt;&lt;code&gt;&lt;span class="gd"&gt;-                   resource.TestCheckResourceAttrPair(&lt;/span&gt;
&lt;span class="gi"&gt;+                   resource.TestMatchResourceAttr(&lt;/span&gt;
                        "data.portkey_scim_workspace_mappings.filtered", "mappings.0.workspace_id",
&lt;span class="gd"&gt;-                       "portkey_workspace.test", "id",&lt;/span&gt;
&lt;span class="gi"&gt;+                       workspaceUUIDPattern,&lt;/span&gt;
                    ),
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/Portkey-AI/terraform-provider-portkey" rel="noopener noreferrer"&gt;Portkey-AI/terraform-provider-portkey&lt;/a&gt;, 8 stars (Nov 3, 2026), Go, MPL-2.0&lt;/li&gt;
&lt;li&gt;Files changed: 1 (test only)&lt;/li&gt;
&lt;li&gt;Lines: +5 / -2&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Oct 6, 2026. Merged: Oct 12, 2026. Time to merge: 5 days, 15 hours&lt;/li&gt;
&lt;li&gt;Review: approved and merged by a maintainer, after a live-API test run&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;The maintainer approved with "Correct fix for #63" and posted a local run against the live API, with &lt;code&gt;TestAccScimWorkspaceMappingsDataSource_withCreatedMapping&lt;/code&gt; passing. They explained why the new assertion still gives real coverage: "the UUID regex adds that the data source reported the API value instead of echoing the slug it filtered on."&lt;/p&gt;

&lt;p&gt;They left two non-blocking nits: add a comment explaining why slug and UUID can't be compared, and inline the package-level regex, since other tests in the package define theirs at the point of use.&lt;/p&gt;

&lt;p&gt;The approval comment was marked "Generated with Claude Code." The live-API test output is separate evidence from the wording of that approval.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Did this change the provider's behavior?
&lt;/h3&gt;

&lt;p&gt;No. Only the test file changed.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/Portkey-AI/terraform-provider-portkey/pull/84" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/Portkey-AI/terraform-provider-portkey/issues/63" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;A test that can never pass trains a team to ignore failures. A small, verified correction brings the signal back.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/portkey-terraform-scim-acceptance-test-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>terraform</category>
      <category>go</category>
      <category>testing</category>
      <category>api</category>
    </item>
    <item>
      <title>Making -no-xattr Actually Skip Extended Attributes in Plakar Backups</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:07 +0000</pubDate>
      <link>https://dev.to/prilog/making-no-xattr-actually-skip-extended-attributes-in-plakar-backups-1161</link>
      <guid>https://dev.to/prilog/making-no-xattr-actually-skip-extended-attributes-in-plakar-backups-1161</guid>
      <description>&lt;p&gt;A flag that was parsed but never passed along, and a test that proves it now reaches the importer.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;Plakar's &lt;code&gt;backup -no-xattr&lt;/code&gt; flag was accepted but never passed to the importer, so Plakar still read extended attributes. On filesystems that don't support them, such as Proxmox's FUSE-mounted &lt;code&gt;/etc/pve&lt;/code&gt;, files failed with "operation not supported". The fix copies the flag into the importer options and adds a test that checks it arrives. It was merged as &lt;a href="https://github.com/PlakarKorp/plakar/pull/2327" rel="noopener noreferrer"&gt;PR #2327&lt;/a&gt; and closed &lt;a href="https://github.com/PlakarKorp/plakar/issues/2045" rel="noopener noreferrer"&gt;issue #2045&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/PlakarKorp/plakar" rel="noopener noreferrer"&gt;Plakar&lt;/a&gt; is an open-source backup tool written in Go. Issue #2045 showed a backup of a Proxmox server's config directory:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;plakar at @s3-ik backup -no-xattr -tag pve /etc/pve
...
        import: &lt;span class="nv"&gt;nodes&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;273/273, &lt;span class="nv"&gt;objects&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;1659/1659, &lt;span class="nv"&gt;errors&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;83
...
✘ /etc/pve/notifications.cfg: operation not supported
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The user had already passed &lt;code&gt;-no-xattr&lt;/code&gt;, and the errors still appeared.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="n"&gt;importerOpts&lt;/span&gt; &lt;span class="o"&gt;:=&lt;/span&gt; &lt;span class="n"&gt;ctx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;ImporterOpts&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;importerOpts&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Excludes&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;cmd&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Excludes&lt;/span&gt;
&lt;span class="n"&gt;importerOpts&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;NoXattr&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;cmd&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;NoXattr&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The regression test registers a probe importer that records the &lt;code&gt;NoXattr&lt;/code&gt; option it receives, then runs a backup with and without &lt;code&gt;-no-xattr&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight go"&gt;&lt;code&gt;&lt;span class="n"&gt;require&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Equal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;t&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;tc&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;want&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;xattrProbeNoXattr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Load&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"-no-xattr must reach the importer so it does not read extended attributes"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/PlakarKorp/plakar" rel="noopener noreferrer"&gt;PlakarKorp/plakar&lt;/a&gt;, 2,102 stars (Nov 3, 2026), Go, ISC&lt;/li&gt;
&lt;li&gt;Files changed: 2 (1 source file, 1 test file)&lt;/li&gt;
&lt;li&gt;Lines: +45 / -0&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened: Aug 21, 2026. Merged: Sep 22, 2026. Time to merge: 31 days, 19 hours&lt;/li&gt;
&lt;li&gt;Review: 3 approvals&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How the review went
&lt;/h2&gt;

&lt;p&gt;Three reviewers approved the change over about three weeks, the last two with "lgtm" and "LGTM thx." It was merged on Sep 22.&lt;/p&gt;

&lt;p&gt;The contribution guide described an AI-PR normalization workflow using the project's Claude plugin. The Nov 3, 2026 review notes did not establish whether that workflow was followed; the approvals and merge do not establish it.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  What did the bug affect?
&lt;/h3&gt;

&lt;p&gt;Any backup where &lt;code&gt;-no-xattr&lt;/code&gt; was meant to avoid reading extended attributes. The issue's example was Proxmox's &lt;code&gt;/etc/pve&lt;/code&gt;.&lt;/p&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/PlakarKorp/plakar/pull/2327" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/PlakarKorp/plakar/issues/2045" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;The production change is one line. The test is what makes it last, because it fails the moment the flag stops reaching the importer again.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/plakar-no-xattr-flag-backup-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>go</category>
      <category>devops</category>
      <category>testing</category>
      <category>opensource</category>
    </item>
    <item>
      <title>Fixing Pezzo's ClickHouse Timestamp Parse Error With One Insert Setting</title>
      <dc:creator>Robert Markovski</dc:creator>
      <pubDate>Tue, 17 Nov 2026 13:54:00 +0000</pubDate>
      <link>https://dev.to/prilog/fixing-pezzos-clickhouse-timestamp-parse-error-with-one-insert-setting-3eb2</link>
      <guid>https://dev.to/prilog/fixing-pezzos-clickhouse-timestamp-parse-error-with-one-insert-setting-3eb2</guid>
      <description>&lt;p&gt;Timestamps in one format, a database expecting another.&lt;/p&gt;

&lt;h2&gt;
  
  
  Short answer
&lt;/h2&gt;

&lt;p&gt;Pezzo's reporting service inserted ISO-8601 timestamps, with milliseconds and a &lt;code&gt;Z&lt;/code&gt;, into ClickHouse &lt;code&gt;DateTime&lt;/code&gt; columns. With ClickHouse's &lt;code&gt;basic&lt;/code&gt; input format, those values fail to parse, and testing a prompt locally raised &lt;code&gt;CANNOT_PARSE_INPUT_ASSERTION_FAILED&lt;/code&gt;. The fix sends the insert with &lt;code&gt;date_time_input_format: "best_effort"&lt;/code&gt;. It was merged as &lt;a href="https://github.com/pezzolabs/pezzo/pull/357" rel="noopener noreferrer"&gt;PR #357&lt;/a&gt; and closed &lt;a href="https://github.com/pezzolabs/pezzo/issues/320" rel="noopener noreferrer"&gt;issue #320&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem
&lt;/h2&gt;

&lt;p&gt;Issue #320, on Pezzo 0.8.0 with Docker Compose, showed the error when running any prompt test:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;ClickHouseError: Cannot parse input: expected '"' before: '.307Z",...': (while reading the value of key timestamp)
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  The fix
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight typescript"&gt;&lt;code&gt;         &lt;span class="n"&gt;format&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"JSONEachRow"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
         &lt;span class="n"&gt;table&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"reports"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
         &lt;span class="n"&gt;values&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;reportToSave&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;        &lt;span class="n"&gt;clickhouse_settings&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;          &lt;span class="n"&gt;date_time_input_format&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"best_effort"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="o"&gt;+&lt;/span&gt;        &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Setting it per insert means the parse works regardless of the server's default.&lt;/p&gt;

&lt;h2&gt;
  
  
  By the numbers
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Repository: &lt;a href="https://github.com/pezzolabs/pezzo" rel="noopener noreferrer"&gt;pezzolabs/pezzo&lt;/a&gt;, 3,279 stars (Nov 3, 2026), TypeScript, Apache-2.0&lt;/li&gt;
&lt;li&gt;Files changed: 1&lt;/li&gt;
&lt;li&gt;Lines: +3 / -0&lt;/li&gt;
&lt;li&gt;Commits: 1&lt;/li&gt;
&lt;li&gt;Opened and merged: Aug 21, 2026. Time to merge: 3 hours, 38 minutes&lt;/li&gt;
&lt;li&gt;Review: merged by a maintainer&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Where can I read more?
&lt;/h3&gt;

&lt;p&gt;See the &lt;a href="https://github.com/pezzolabs/pezzo/pull/357" rel="noopener noreferrer"&gt;pull request&lt;/a&gt; and the &lt;a href="https://github.com/pezzolabs/pezzo/issues/320" rel="noopener noreferrer"&gt;original issue&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  The goal
&lt;/h2&gt;

&lt;p&gt;Observability data is only useful if it lands. Telling the database how to read the timestamps it is sent keeps the reports flowing.&lt;/p&gt;




&lt;p&gt;This fix was developed with &lt;a href="https://prilog.ai" rel="noopener noreferrer"&gt;Prilog&lt;/a&gt;. Read the original case study on the &lt;a href="https://prilog.ai/blogs/pezzo-clickhouse-datetime-best-effort-fix/" rel="noopener noreferrer"&gt;Prilog blog&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>clickhouse</category>
      <category>typescript</category>
      <category>observability</category>
      <category>debugging</category>
    </item>
  </channel>
</rss>
