Move `docs/defensive-patterns.md` into `docs/defensive-patterns/`, one page per
artifact plus a README router, and state each rule as its requirement, the
condition a reviewer judges it by, its boundary and approver, and a correct and
a wrong example.
- `general.md` (the moved page): the existing bug-class rules plus timing,
value representation, shared-state ownership, decentralization, performance
evidence, and migrations.
- `package-surface.md`, `remote-and-api.md`, `client-components.md`,
`conversation-and-session.md`: the review requirements for those artifacts,
linking type definitions and semantics to `docs/subsystems/` instead of
restating them.
- `README.md`: the habits behind the rules and the route from artifact to page.
- `dsh-code-review` reads the owning pages before judging a diff; the
`dsh-architecture-discipline` routing skill is retired in favour of the
directory, and every inbound reference is repaired.
- Chinese counterparts, pairing records, and the documentation tier table are
updated together.
A dismissed review now returns a successful exit_plan_mode result, so
the plan policy must not treat any success as leaving plan mode. Also
update the user-facing description of a dismissed review.
Silero cannot back-date the first segment of a cold detector, so it
reports a start index several frames after speech actually began. Every
recording therefore lost its opening audio: measured across 50
utterances the median loss was 102 ms and the worst 230 ms — a full
syllable, which surfaced as dropped leading words ("麻烦把发票…" →
"把发票…", "用 Python…" → "写一个 Python…").
Only the first segment is affected; later segments aligned within 3 ms
because the detector retains history by then. Tuning cannot fix it:
lowering minSpeechSeconds from 0.25 s to 0.05 s left the median loss at
110 ms, and lowering the threshold to 0.2 still lost up to 174 ms while
risking false triggers.
Restore the dropped onset from the recording itself, bounded by the new
vadOnsetPaddingSeconds (default 0.3 s, zero disables). Measured on the
same corpus with identical DSH settings, this recovers the full no-VAD
accuracy and in the technical domain exceeds it, since segmentation also
trims silence:
语料 条件 无VAD VAD丢起音 VAD+回填 改善
audio clean 12.50% 16.33% 12.50% 23.5%
audio noisy10 17.74% 19.15% 17.94% 6.3%
audio_tech clean 10.34% 15.65% 9.81% 37.3%
audio_tech noisy10 25.73% 30.50% 26.26% 13.9%
Inference cost is unchanged (2.6 s vs 2.6 s on both clean corpora).
Extension failures, unreadable package identities, a blocked session-log
upload, and degraded Messages replay recur on every request. Log the first
occurrence per field, route, module, or blocking event instead.
Co-authored-by: turtle1999 <[email protected]>
Request-extension fields are auxiliary. A contributor whose preparation
throws is now logged and omitted while the other fields are still sent,
and a post-2xx acceptance failure is logged without failing the request.
Package inventory is best effort: a named manifest without a version, or
a bare package whose manifest is missing or unreadable, reports version
"unknown" instead of failing preparation.
imccyu