Cx Dev Log — 2026-04-16

#cx #programming

No developer commits landed anywhere in the repo today. This is the third consecutive day of zero human activity. The matrix holds at 78/78 on main, the working tree is clean, and the only recent commits are automated documentation entries.

That said, the state of the project is worth documenting because the integration debt is growing in a way that matters.

The submain gap

submain sits 15 work commits ahead of main, carrying +1,974 lines across 97 files. This includes four hard-blocker closures that landed between March 31 and April 13:

Basic test runner
Minimal error model (Result, Ok, Err, ?)
Integer overflow enforcement at declared width
Optional semicolons

There is also a full parser/semantic/interpreter audit (Part 1), diagnostics readability improvements, Phase 10 loop lowering (for/while/loop/break/continue), and wrapping arithmetic enforcement. All of this is committed and working on submain but not yet on main.

The merge has been the predicted next step in every daily log since April 13. It has not happened.

Branch backlog

The daily-log branch backlog grew to 14 unmerged branches today. Each carries a single commit against main. The site branch is 15 blog posts ahead of main. None of this is blocking anything functionally, but it widens the conflict surface slightly each day that the submain merge doesn't happen.

What is actually blocking the merge

Probably nothing technical. The one known friction point is a roadmap file conflict: docs/frontend/ROADMAP.md is at v4.8 on main (from March 28) while submain carries edits against v4.7. That is a straightforward text conflict to resolve manually. The 78/78 matrix pass rate on main suggests no regression risk from the merge itself.

What remains on the critical path

These items are unchanged from the past three logs:

Merge submain into main. This unblocks everything downstream and closes four hard blockers on the canonical branch.
Fix audit_09: compound assignment on struct fields skips width truncation. One function change in run_semantic_stmt.
Convert result_type to recursive() to unlock nested generic composition.
Implement Result in IR lowering. The interpreter path works on submain; lower.rs still has unsupported! arms for the Result-related constructs.
Audit Part 2: arrays, imports + generics, nested generics, copy_into under Result.

Looking ahead

The project is in a holding pattern. Fifteen commits of real engineering work are waiting on a single merge operation. The longer the gap persists, the more the daily-log and site branches diverge from main, making the eventual reconciliation slightly messier. The actual merge itself should be low-risk given the passing test matrix, but it needs a human to initiate it.

Follow the Cx language project: