Cx Dev Log — 2026-05-26

#cx #programming

Stuck without a single developer-authored commit landing today, while submain sits 7 commits ahead of main, full of crucial work waiting to merge. The backlog's growing increasingly hard to ignore.

Submain is sitting on real work

Submain's ahead, and it's not just sitting there with trivial updates. We've got 7 commits affecting 47 files: a whopping +1061/-331 lines. Inside? Real changes like:

Tracker #001: Tackling composite literal type-checking.
Tracker #002 / #012 / #032: Nailing down IndexOutOfBounds diagnostics.
Tracker #008: The Builtin registry.
Tracker #009: A killer 57% speedup on arithmetic loops using BindingId-indexed variables.
Implementation of the exit() builtin.

Yet, none of this has hit main. No merge conflicts are lurking ― submain was last synced with main on May 24 through PR #282. Each daily log promises a merge from submain to main since May 6, and every promise has fizzled out.

The daily-log branch backlog

The issue is bigger than just submain. The daily-log branches are stacking up too. The last time main saw a daily-log merge was back on 2026-05-09. Branches from 05-24 and 05-25 are just gathering dust. They’ve got crucial roadmap updates and project history waiting in the wings.

Right now, main’s roadmap still reads "0.1 release candidate" with a 117/117 matrix. Reality check: we're at v0.1.0 with 182 tests passing. Daily-log branches know this, but without merging, it's nowhere on the main branch.

Nothing in progress

Main’s working tree is clean. No files in limbo or half-baked changes. It’s not that the work’s just somewhere else. It’s a genuine pause.

What is likely next

The crystal ball looks a lot like yesterday's prediction: submain needs to merge into main. That’s the most impactful move we've got. Following that, merging daily-log branches would pull the documentation up to date. Work branching off from the May 24-25 tracker sprint is all about either optimizing lazy arenas (tracker #010, targeting that annoying 64KB memset bottleneck) or tightening JIT array bounds (tracker #003, nailing the jit_known_unsound issues).

Whether any of this unfolds tomorrow remains to be seen. But the work's definitely there, waiting for the green light.

Follow the Cx language project: