lint-once: run: >- block scalars are caught only by a test whose name does not say so #35

New Issue

clawbot · 2026-09-04T14:12:22+02:00

clawbot commented

2026-09-04 14:12:22 +02:00

Spun out of #31 under its final-round rule: found during review of that PR, recorded rather than fixed there.

A run: >- (or run: |) step in .gitea/workflows/check.yml resolves to the bare scalar indicator, which reaches nothing. The prettier count therefore stays 1 no matter what the block contains — what actually turns such a step red is the exact-equality assertion in it("reads the run steps of the CI workflow and not its uses steps").

That works, and #31 corrected the header where it had claimed otherwise. The residue is that the guard lives in a test whose name gives no hint it is load-bearing for block scalars, and its cost is real: any legitimate second run: step — a cache step, an echo — fails that test for a reason unrelated to prettier. A comment in the file says so.

Not a false green and not urgent. Options: resolve block scalars properly and follow their contents, or move the guard into a test named for what it guards. Either way the file's comments must end up matching the code exactly, which is the standard #31 established.

PR #31 for context.

Spun out of #31 under its final-round rule: found during review of that PR, recorded rather than fixed there. A `run: >-` (or `run: |`) step in `.gitea/workflows/check.yml` resolves to the bare scalar indicator, which reaches nothing. The prettier **count** therefore stays 1 no matter what the block contains — what actually turns such a step red is the exact-equality assertion in `it("reads the run steps of the CI workflow and not its uses steps")`. That works, and #31 corrected the header where it had claimed otherwise. The residue is that the guard lives in a test whose name gives no hint it is load-bearing for block scalars, and its cost is real: any legitimate second `run:` step — a cache step, an `echo` — fails that test for a reason unrelated to prettier. A comment in the file says so. Not a false green and not urgent. Options: resolve block scalars properly and follow their contents, or move the guard into a test named for what it guards. Either way the file's comments must end up matching the code exactly, which is the standard #31 established. PR #31 for context.

~~clawbot referenced this issue 2026-09-04 14:12:41 +02:00~~

Run all linting in Docker via Dockerfile.lint (closes #30) #31

clawbot commented

2026-09-05 11:32:54 +02:00

Closed at sneak's instruction, 2026-09-05: this was opened by an agent running on another machine outside the managed fleet, under superseded rules. Reopen if the content is wanted.

Model: fable-5-1

Closed at sneak's instruction, 2026-09-05: this was opened by an agent running on another machine outside the managed fleet, under superseded rules. Reopen if the content is wanted. Model: fable-5-1

clawbot closed this issue

2026-09-05 11:32:54 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: sneak/quak#35

lint-once: run: >- block scalars are caught only by a test whose name does not say so #35

lint-once: `run: >-` block scalars are caught only by a test whose name does not say so #35