StoneCypher / jssm / 4aefe322fe9238f33854fc5ee88c7fa2d0f900fe

100%

master: 100%

Pull #973

github

docs(specs): correct the highlightjs CI claim; reopen the scoped-name question

The previous commit asserted "no grammar repo in the org runs the
language tests in CI. Not one." That is FALSE. It was drawn from file
listings without reading two workflows whose names should have been
obvious tells.

Surveyed properly, 2 of 8 run their language tests on every push and PR:

- highlightjs-solidity     test.yml, "Build and Test", npm ci -> npm test
- highlightjs-cshtml-razor node.js.yml, matrix Node 10/12/14, npm test

Both use their OWN mocha runner with highlight.js as a devDependency,
not the core extra/ harness -- which is the actually useful finding: a
repo that wants CI writes its own runner, because the documented
workflow needs a checkout of another project and does not fit an Action.

The org is actively maintained, not dormant: highlight.js pushed
2026-07-26, apex 2026-07-22, jai 2026-07-21.

What survives: the .expect.txt / .expected.txt hazard is still real, and
highlightjs-luau still ships both spellings in one directory -- luau
being one of the three repos with no CI at all, which is why nobody
caught it. Narrower lesson, same direction: write a runner and wire it
to CI, following solidity and cshtml-razor.

Also reopens 2a. The scoped-name rejection rested on the org reading as
a loose collection of drive-by grammars; an actively maintained org
changes that calculus. Marked UNRESOLVED rather than flipped. The
publish-unscoped-first ordering holds either way, since the unscoped
name is an unclaimed squat target today.

Claude-Session: https://claude.ai/code/session_01EBLDd3nmRPcpC2weqD2TD3

3869 of 3869 branches covered (100.0%)

Branch coverage included in aggregate %.

6361 of 6361 relevant lines covered (100.0%)

33379.13 hits per line