humanspeak / svelte-virtual-chat / 29059991150

92%

main: 92%

push

github

chore(lint): lint unit test files and enable type-aware promise rules (#57)

* chore(lint): lint unit test files and enable type-aware promise rules

Remove the '**/*.test.ts' ignore so the 7 vitest unit-test files are
statically checked, and swap ts.configs.recommended for
recommendedTypeChecked with projectService so type-aware rules
(no-floating-promises, no-misused-promises) receive type information.

- Add a scoped test override relaxing no-non-null-assertion
- Add disableTypeChecked blocks for root config files and docs/
- Resolve violations surfaced by the newly-enabled typed rules

Snapshot committed by guard for review against
.agents/.plans/typed-lint/001-lint-unit-tests-and-typed-lint.md

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* docs(guard): record checkpoint 1 and NO-PASS close-out for typed-lint plan 001

Guard's own record of the final gate on 4da326c, kept separate from the
source snapshot it reviews.

Verdict NO-PASS. Typed linting is genuinely wired — no-floating-promises
provably fires in .ts, .test.ts and .svelte — and every gate is green.
But no executor report was written, so three done criteria that depend on
it fail, and the plan's named .svelte example (stream-swap:228) turns out
to be uncatchable: checksVoidReturn.attributes is JSX-only and cannot see
Svelte event attributes.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* docs(guard): amend typed-lint plan 001 to remove a false coverage claim

Checkpoint 1 proved Step 2.5's central prediction wrong: no-misused-promises'
checksVoidReturn.attributes inspects JSX attributes only, so a promise-returning
Svelte event handler (stream-swap/+page.svelte:228) is unreachable by the rule.
Left as written, the plan would keep directing executors to "fix" a
non-violation and leave readers believing Svelte event handlers are
promise-safe.

Amended with the operator's explicit agreement:

- Add a dated revision note under the executor-instructions block
- Replace... (continued)

216 of 248 branches covered (87.1%)

Branch coverage included in aggregate %.

3 of 3 new or added lines in 1 file covered. (100.0%)

392 of 410 relevant lines covered (95.61%)

89.8 hits per line