test(#4518): migrate loop/hook-point e2e batch to named timeout constants

Batch 7 of 17 in the ad hoc timeout literal migration (epic #4445).
Replaces every bare numeric timeout/timeoutMs object-literal property in
tests/loop-render-hooks.test.cjs, tests/loop-walk.qa.test.cjs,
tests/loop-hooks-empty-points-e2e.test.cjs, tests/loop-hooks-ship-pre-e2e.test.cjs,
tests/loop-hooks-verify-post-e2e.test.cjs, tests/check-gap-analysis-plan-post-e2e.test.cjs,
tests/check-tdd-review-checkpoint-e2e.test.cjs, tests/execute-wave-post-gate-pipeline-e2e.test.cjs,
tests/plan-pre-hook-e2e.test.cjs, and tests/qa/tdd-walk.cjs with a named
constant, per eslint-rules/no-adhoc-timeout-literal.cjs. Removes the 10
files from the rule's allowlist.

Promotes a new shared class norm to tests/helpers/timeouts.cjs,
LOOP_HOOK_POINT_CLI_TIMEOUT_MS: 7 files independently arrived at the same
value for a single gsd-tools.cjs CLI subcommand invocation with no
confirmed subprocess fan-out. Reuses the existing PROBE_TIMEOUT_MS for
loop-render-hooks.test.cjs's 11 sites (same class, exact value match).
Adds 4 file-local constants for values that share a class with the new
norm or an existing one but diverge in pre-existing value, or that
numerically coincide with an unrelated existing constant without
matching its actual operation.

Isolated Standards-axis review caught that the new shared constant's doc
comment exhaustively enumerated 3 verb families while a 7th genuine site
(an `init new-project` invocation) also correctly belonged to the class;
fixed by rewording the comment to state the class definition (call
shape) first and list all 4 representative verbs, explicitly
illustrative rather than exhaustive. No value changed, no site's
classification changed.

No src/bin file touched, no numeric value changed anywhere.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
sim
2026-09-10 15:50:52 -04:00
parent 138e70d734
commit 6d6e3eea73
12 changed files with 91 additions and 31 deletions

View File

@@ -159,6 +159,35 @@ const FIXTURE_HOOK_TIMEOUT_SECONDS = 5;
*/
const STAGED_HOOK_SCRIPT_TIMEOUT_MS = 20000;
/**
* A single `gsd-tools.cjs` CLI subcommand invocation, spawned directly (never
* via an intermediate shell script), used by a loop/hook-point e2e test. The
* command does real in-process work -- capability/config resolution, project
* scaffolding, or (for the `check` verbs that inspect git-tracked state) real
* git plumbing calls -- but as ONE process with no confirmed nested-subprocess
* fan-out, verified per-site by reading each file's own runner function, not
* assumed from "hook" in the filename. Representative verbs seen at this
* norm's 7 sites: `loop render-hooks <point>`, `check <check-id>`,
* `execute <hook-point>`, and `init new-project` (tests/loop-walk.qa.test.cjs's
* concurrency test) -- the norm describes the CALL SHAPE (single direct
* gsd-tools.cjs spawn, no fan-out), not one specific verb family, so this list
* is illustrative, not exhaustive.
*
* Distinct from `HOOK_FANOUT_TIMEOUT_MS` (a bash-hosted script that itself
* shells out to further subprocesses) despite the coincidentally-matching
* value: this norm's sites spawn `gsd-tools.cjs` and nothing beneath it.
* Distinct from `PROBE_TIMEOUT_MS`: the same underlying `loop render-hooks`
* command is ALSO called at `PROBE_TIMEOUT_MS` (15000ms) elsewhere in the
* suite against lighter temp-project fixtures -- the two bounds are kept
* separate rather than equalized, since this migration never raises or
* lowers a value without a fresh bench citation.
*
* Shared across 7 files in batch #4518 of the ad hoc timeout literal
* migration, epic #4445 -- every site independently arrived at this exact
* value -- that is why it lives here rather than as a file-local constant.
*/
const LOOP_HOOK_POINT_CLI_TIMEOUT_MS = 60000;
module.exports = {
PROBE_TIMEOUT_MS,
HOOK_FANOUT_TIMEOUT_MS,
@@ -170,4 +199,5 @@ module.exports = {
QUICK_SPAWN_TIMEOUT_MS,
FIXTURE_HOOK_TIMEOUT_SECONDS,
STAGED_HOOK_SCRIPT_TIMEOUT_MS,
LOOP_HOOK_POINT_CLI_TIMEOUT_MS,
};