Tom Boucher 325fc25c01 fix(#3569): require a digit-bearing phase id in the stats heading scan (#3591) ...

* test(#3569): pin stats phase-id shape — inline-code mentions produce no phantom row

Failing-first regression for #3569: cmdStats' heading scan accepted any
word as a phase id, so prose mentioning ### Phase N: inside inline code
inflated phases_total and disagreed with roadmap analyze. New adversarial
fixture phase-heading-inside-inline-code.md (blockquote + bare mention),
parity assertion against roadmap analyze, and over-narrowing guards for
decimal / milestone-prefixed / letter-prefixed ids.

* fix(#3569): require a digit-bearing phase id in the stats heading scan

cmdStats' hand-rolled heading pattern captured any word as a phase id,
so a ### Phase N: token inside an inline code span (the issue's
blockquote) produced a phantom Not-Started row that could never
complete, inflating phases_total and deflating percent forever. The id
capture is now the canonical #3036 shape roadmap.cts uses (digit
required; letter-prefixed, decimal, and milestone-prefixed ids keep
counting), so stats and roadmap analyze agree.

* fix(#3569): sanction the stats id-shape literal; correct zero-padded expectation

Review findings: the phase-id drift guard requires the // phase-id-owner:
comment directly above the regex (same form as roadmap.cts); the
milestone-prefixed over-narrowing guard must expect normalizePhaseName's
zero-padded 02-01 form, not the raw 2-01 token.

* chore(#3569): add changeset fragment

* chore(#3569): backfill changeset pr number

---------

Co-authored-by: sim <sim@local>

2026-08-17 11:39:01 -04:00

4.4 KiB

Raw Blame History

View Raw