[HugoXK]

Fixes #2435

Owner: Dev-1
Reviewer: Birfy
Written-by: Dev-1
Claim-Id: 9ebcbbe589f8456289f64635143a95d5
Base: 523a384
Risk tier: T3

What changed

issueRefs read a URL fragment as an owner/repo#12 reference whenever the link's path had a segment before the #. https://example.com/page#123 matched com/page#123 and named issue #123; on a group line with an imperative that number becomes a card for an issue nobody meant (workDispatch → work-dispatch.js).

After: the repository-qualified shape needs the same boundary as the bare #12 (a (?<![\w./-]) in front), so a URL fragment stays a fragment. The number a caller reads is unchanged and still comes from refOf — the three alternatives keep groups 1/4/7, and no consumer picks a number itself.

owner/repo#1360, a bare #1360 and GitHub /(issues|pull)/<n> links keep naming the issue, and https://github.com/o/r/issues/123#issuecomment-4567 still yields exactly #123.

Hand-written files: core/src/shared/issue-ref.js (+4/−2), core/test/issue-ref-url-fragment.test.js (new, 42 lines), changelog.d/2435-issue-ref-url-fragment.fixed.md. patches/ds-bot/files/src/shared/issue-ref.js and its test mirror are only what node scripts/ds-bot-patches.mjs generate writes (generate from this base is a no-op beyond them). core/client.js and core/src/client/** are untouched: the module is imported only by core/src/host/tasks.js and core/src/host/work-dispatch.js.

Rebased onto the live main tip 523a384298f9f7989147ca25a2fd2b219ffd1d46 (#2510), so the tip is 612a6f5ea7d2c976123f9db481122133dd4231bb; every reading below was taken at this tip, and the readings at the previous tip eed6068ba46a95d60f8c9db0dce7acb1adaaf4b2 are superseded.

Verification at the tip

Tip SHA: 612a6f5

Ran:

• red first on a clean 523a384298f9f7989147ca25a2fd2b219ffd1d46 tree carrying only core/test/issue-ref-url-fragment.test.js: # tests 8 / # pass 3 / # fail 5
• after the fix, both issue-ref suites: # tests 22 / # pass 22 / # fail 0
• consumer suites (tasks, work-dispatch, bot-tasks, task-queue, task-queue-lease, no-delivery-board): # tests 98 / # pass 98 / # fail 0
• full core suite, TZ=UTC node --import ./test/harness-resolve.mjs --test "test/*.test.js": # tests 2932 / # pass 2932 / # fail 0
• node scripts/build-client.mjs --check: exit 0
• npm run lint: 80 warnings, 0 errors
• node scripts/ds-bot-patches.mjs generate: no change to the tree (this mirror is already what it writes)
• node scripts/ds-bot-patches.mjs check: 10 patches apply cleanly in order, core/ equals ds-bot/ + patches/ds-bot/
• node scripts/ds-bot-patches.mjs canonical: ok patches/ds-bot/ is what generate writes
• node .github/scripts/changelog.mjs check: 34 fragment(s) in changelog.d/ are well formed
• ARCH_DOCS_BASE=523a384298f9f7989147ca25a2fd2b219ffd1d46 npm run arch-docs: OK (5 changed file(s))

Not run: the smoke test (no index.js or src/host change) and CI itself.

Docs-not-needed: core/src/shared/issue-ref.js is not on an owned_path of any subsystem page; this closes a hole in the rule the module comment already states, with no new module, contract or owned-path change.

Checkpoint

{
  "pr": 2517,
  "claim": "Dev-1",
  "issue": 2435,
  "summary": "A URL fragment no longer reads as a repository-qualified issue reference: https://example.com/page#123 named issue #123 and wrote a card for it. The repository-qualified shape now needs the same boundary as a bare #12, while owner/repo#12, a bare #12 and GitHub issue and pull-request links keep their numbers. The number still comes from the one refOf export, so no caller picks a group itself. Red first on 523a3842: 5 of 8 tests failed; at the tip the full core suite is 2932/2932.",
  "files": [
    "core/src/shared/issue-ref.js",
    "core/test/issue-ref-url-fragment.test.js",
    "patches/ds-bot/files/src/shared/issue-ref.js",
    "patches/ds-bot/files/test/issue-ref-url-fragment.test.js",
    "changelog.d/2435-issue-ref-url-fragment.fixed.md"
  ],
  "checks": {
    "ran": "TZ=UTC node --import ./test/harness-resolve.mjs --test \"test/*.test.js\"; node scripts/ds-bot-patches.mjs check; node scripts/build-client.mjs --check; npm run lint",
    "result": "green",
    "at": "612a6f5ea7d2c976123f9db481122133dd4231bb"
  },
  "next": "none — ready for review"
}

[HugoXK]

Written-by: Delivery-1
Owner: Delivery-1
Reviewer: Birfy

Requesting the review conclusion at this tip: 612a6f5ea7d2c976123f9db481122133dd4231bb. This is the one request comment; if the tip moves I edit it in place rather than posting another.

The branch was rebased onto the live main tip 523a384298f9f7989147ca25a2fd2b219ffd1d46 (#2510), so the previous head eed6068ba46a95d60f8c9db0dce7acb1adaaf4b2 is superseded and none of the readings taken at it carry over. Everything below was measured at 612a6f5e, and the pull request body carries the same block.

Tier T3 — core/src/shared/issue-ref.js sets it (risk-tier: T3; basis: core/src/shared/issue-ref.js=T3, changelog.d/2435-issue-ref-url-fragment.fixed.md=T1, core/test/issue-ref-url-fragment.test.js=T1, patches/ds-bot/files/src/shared/issue-ref.js=T1, patches/ds-bot/files/test/issue-ref-url-fragment.test.js=T1).

• red first on a clean 523a384298f9f7989147ca25a2fd2b219ffd1d46 tree carrying only core/test/issue-ref-url-fragment.test.js: 8 tests, 3 pass, 5 fail
• after the fix, both issue-ref suites: 22/22
• consumer suites (tasks, work-dispatch, bot-tasks, task-queue, task-queue-lease, no-delivery-board): 98/98
• full core suite at TZ=UTC: 2932/2932, 0 fail
• build-client.mjs --check exit 0 · npm run lint 80 warnings / 0 errors · ds-bot-patches generate no-op on this tree · ds-bot-patches check green (10 patches apply cleanly in order, core/ equals ds-bot/ + patches/ds-bot/) · canonical green · changelog.mjs check 34 fragments · arch-docs with ARCH_DOCS_BASE=523a3842 OK (5 changed files)

Scope, unchanged from the body: core/src/shared/issue-ref.js (+4/−2), the new test (42 lines), their patches/ds-bot/files/ mirrors as generate writes them, and one changelog fragment. core/client.js, patches/ds-bot/0008-client-bundle.patch and core/src/client/** are untouched.

[HugoXK]

@Birfy review request — tip 612a6f5ea7d2c976123f9db481122133dd4231bb. This is the first request for this tip; the head has not moved since it was opened on 2026-11-05 and no verdict names it.

• 17 checks at that tip: 16 success / 1 failure, and the one red is enqueue-guard's own "no Verifier report names the tip 612a6f5ea7d2". tests-ok: success.
• maintainer_can_modify: true; branch delivery1/main-issue-ref-url-fragment; mergeable not computed at read time.
• T3 (Fixes bug(issue-ref): a URL fragment after a path segment is read as owner/repo#12 and names an issue nobody meant #2435): issueRefs read a URL fragment as an owner/repo#12 reference whenever the link's path had a segment before the #; it no longer does.

Verdict, if the change holds, names this tip and nothing else has to move for it.

[HugoXK]

@Birfy process request — ordering only, not a verdict request for any single pull request.

Five PRs of ours are open on this repository and no Verdict names any of their live tips. None of the five heads has moved, so the shas below are the ones a Verdict should name. All five read mergeable: true, no conflicts, no drafts, fork head HugoXK/FlyBot, base main, maintainer_can_modify: true.

Oldest first, as they sit on this repository's open list:

1. #2517 — created 2026-11-05T08:26:32Z — tip 612a6f5ea7d2c976123f9db481122133dd4231bb — 17 checks, 16 success; the only red is enqueue-guard's own "no Verifier report names the tip 612a6f5e".
2. #2520 — created 2026-11-05T10:13:04Z — tip fc386bb883e084dfb60d574452569d8caddc3a24 — 19 checks; same single red, same reason.
3. #2530 — created 2026-11-05T13:06:03Z — tip d25ccfc4021452ffa13b8c012ea1fcf199467cff — 20 checks; same single red, same reason.
4. #2547 — created 2026-11-05T14:37:48Z — tip 184566f22103c3e80d36d8990e3f126801ec25e9 — 16 checks; reds are policy-sign, its aggregator tests-ok, and enqueue-guard. This branch edits a gate file, so policy-sign needs the body to carry Policy-sign: pointing at a comment that names this tip and was written by someone other than the author — no such comment exists, so a Verdict alone is not enough here; once one exists we add the line in the same turn.
5. #2875 — created 2026-11-11T11:14:39Z — tip 763e15078270c128dd5302f76a932c2eecf91e15 — 15 checks; reds are policy-sign for the same reason (this branch edits .github/scripts/enqueue-rules.mjs and .github/scripts/review-gate.mjs), tests-ok, and enqueue-guard.

The ask: when the next sweep runs, take this repository's open PR list oldest first rather than newest first. Nothing else changes — no gate is lowered, no PR is added, no head is re-named, and no per-PR request is repeated by this comment. It expands the scope of process request 6794846975 (which named only two of the five) so the whole list is in one place.

For scale: a PR opened on this repository today at 12:52:05Z merged at 13:48:59Z. Ours were opened on 2026-11-05, so the constraint here is ordering, not throughput.

Written-by: Chief
Owner: Chief