Cx Dev Log — 2026-05-27

#cx #programming

Three quiet days without any developer-authored commits. That's where we stand, with no changes to main since May 22 when the Type::Void variant merged in PR #278. Meanwhile, submain sits seven commits ahead, echoing the unmerged work from the recent tracker sprint. The passing matrix at 182 and zero failures is a testament to recent progress. But while it may seem like a standstill, the project is poised for a leap forward as integration nears.

Roadmap corrections

There's been a backlog of housekeeping, affecting our roadmap's clarity. Stuck on a May 9 version, it wrongly labeled the frontend as a "0.1 release candidate" and maintained an outdated "117/117" test pass count, despite 182 tests now passing. Today's log addresses these discrepancies. Key overlooked items in Phase 11 had already made their way to main yet remained unchecked:

Method call lowering (0ab7e9b): Executes literal-width narrowing for method calls on implementation types.
when block lowering (bed71c1): Handles Literal, Range, Bool, and Catchall arms alongside TBool wire-value unknowns.
DotAccess in compound forms (2d6804a): Enables compound assignments through dot-access paths, validated by the CX-187 parity fixtures.

These entries, committed weeks ago, highlight the pitfalls of maintaining documentation external to code changes. Delaying integration to main has hidden completion, but the overview now reflects the present state.

A growing submain backlog

Submain harbors seven pivotal commits from the May 24-25 sprint:

Tracker #001: Composite literal type-checking
Tracker #002 / #032: Improved IndexOutOfBounds diagnostics
Tracker #008: Builtin registry implementations
Tracker #009: Transition to BindingId-indexed variables; results in a substantial 57% speedup on arithmetic loops
Tracker #012: Introduction of the exit() builtin, as first discussed in the May 23 log

With no anticipated merge conflicts—submain already aligned with main via PR #282—the merge is strategic. An action expected for days, pending outcomes and persistent in daily logs since May 6.

The log branch tangle

We've accumulated seven daily-log branches (May 20 through May 26), each carrying roadmap updates and notes. The last log branch to merge was way back on May 9. Corrections in today's log address most of the lingering matters in these intermediate branches. Still, they need merging or closing to alleviate any branch clutter.

On the horizon

The focus remains firmly on these essential tasks:

Merge submain into main. This will incorporate critical structural improvements, performance refactors, and necessary fixes from our last sprint. A clean and essential step.
Deal with the daily-log branch backlog. With seven pending branches, tidying up is non-negotiable. Effective branch hygiene is critical, even if corrections lessen the pressure.
Next issues: Lazy arena (tracker #010) or JIT bounds checking (tracker #003). Deciding which to tackle will follow naturally. The former addresses a 64KB memset limit; the latter closes the jit_known_unsound gap.

The truth is, post-May 25, we've hit a productivity pause. The groundwork exists; we just need to anchor it into place.

Follow the Cx language project: