mirror of
https://github.com/ksyasuda/SubMiner.git
synced 2026-04-11 16:19:27 -07:00
chore(repo): update vendor and backlog tasks
This commit is contained in:
@@ -0,0 +1,44 @@
|
|||||||
|
---
|
||||||
|
id: TASK-287
|
||||||
|
title: Restore Lua parser compatibility for mpv plugin modules
|
||||||
|
status: Done
|
||||||
|
assignee: []
|
||||||
|
created_date: '2026-04-11 21:25'
|
||||||
|
updated_date: '2026-04-11 21:29'
|
||||||
|
labels:
|
||||||
|
- bug
|
||||||
|
- mpv-plugin
|
||||||
|
- lua
|
||||||
|
dependencies: []
|
||||||
|
---
|
||||||
|
|
||||||
|
## Description
|
||||||
|
|
||||||
|
<!-- SECTION:DESCRIPTION:BEGIN -->
|
||||||
|
Users with Lua runtimes that do not accept the current `goto continue` pattern in the mpv plugin should be able to load the plugin without syntax errors. Remove the parser-incompatible control-flow usage from the affected plugin modules without changing plugin behavior.
|
||||||
|
<!-- SECTION:DESCRIPTION:END -->
|
||||||
|
|
||||||
|
## Acceptance Criteria
|
||||||
|
<!-- AC:BEGIN -->
|
||||||
|
- [x] #1 The mpv plugin source no longer relies on parser-incompatible `goto continue` labels in the affected Lua modules.
|
||||||
|
- [x] #2 Automated coverage fails on the old parser-incompatible source and passes once the compatibility fix is in place.
|
||||||
|
- [x] #3 Existing plugin start/gate verification still passes after the compatibility fix.
|
||||||
|
<!-- AC:END -->
|
||||||
|
|
||||||
|
## Implementation Notes
|
||||||
|
|
||||||
|
<!-- SECTION:NOTES:BEGIN -->
|
||||||
|
Reused existing local cleanups in `plugin/subminer/hover.lua` and `plugin/subminer/environment.lua` to remove `goto continue` / `::continue::` control flow without behavior changes.
|
||||||
|
|
||||||
|
Added `scripts/test-plugin-lua-compat.lua` and wired it into `test:plugin:src`; the regression checks reject the legacy pattern structurally and verify parse success with `luajit` when available.
|
||||||
|
|
||||||
|
Verification run on 2026-04-11: `lua scripts/test-plugin-lua-compat.lua` ✅, `bun run test:plugin:src` ✅, `bun run changelog:lint` ✅, `bun run typecheck` ✅, `bun run test:env` ✅, `bun run build` ✅, `bun run test:smoke:dist` ✅.
|
||||||
|
|
||||||
|
`bun run test:fast` remains red for unrelated existing immersion-tracker assertions in `src/core/services/immersion-tracker/__tests__/query-split-modules.test.ts` and `src/core/services/immersion-tracker/__tests__/query.test.ts` (`tsMs`/`lastWatchedMs` observed as `-2147483648`).
|
||||||
|
<!-- SECTION:NOTES:END -->
|
||||||
|
|
||||||
|
## Final Summary
|
||||||
|
|
||||||
|
<!-- SECTION:FINAL_SUMMARY:BEGIN -->
|
||||||
|
Removed parser-incompatible `goto continue` usage from the affected mpv Lua plugin modules, added a dedicated Lua compatibility regression script to the plugin test lane, and added a changelog fragment for the user-visible fix. Requested plugin verification is green; unrelated existing `test:fast` immersion-tracker failures remain outside this task.
|
||||||
|
<!-- SECTION:FINAL_SUMMARY:END -->
|
||||||
@@ -0,0 +1,42 @@
|
|||||||
|
---
|
||||||
|
id: TASK-288
|
||||||
|
title: Stabilize immersion-tracker CI timestamp handling under libsql/Bun
|
||||||
|
status: Done
|
||||||
|
assignee: []
|
||||||
|
created_date: '2026-04-11 21:34'
|
||||||
|
updated_date: '2026-04-11 21:43'
|
||||||
|
labels:
|
||||||
|
- bug
|
||||||
|
- ci
|
||||||
|
- immersion-tracker
|
||||||
|
dependencies: []
|
||||||
|
---
|
||||||
|
|
||||||
|
## Description
|
||||||
|
|
||||||
|
<!-- SECTION:DESCRIPTION:BEGIN -->
|
||||||
|
`bun run test:fast` is currently failing because large millisecond timestamps are not handled safely through the libsql/Bun path. Fix timestamp parsing/storage so lifetime/library and session-event queries return correct wall-clock values in CI and runtime.
|
||||||
|
<!-- SECTION:DESCRIPTION:END -->
|
||||||
|
|
||||||
|
## Acceptance Criteria
|
||||||
|
<!-- AC:BEGIN -->
|
||||||
|
- [x] #1 Large wall-clock timestamps round-trip correctly through immersion-tracker lifetime/library queries under the repo's Bun/libsql runtime.
|
||||||
|
- [x] #2 Session-event timestamps round-trip correctly for real wall-clock values used by runtime event inserts.
|
||||||
|
- [x] #3 Targeted immersion-tracker regression coverage passes, and the previously failing `test:fast` lane no longer fails on these timestamp assertions.
|
||||||
|
<!-- AC:END -->
|
||||||
|
|
||||||
|
## Implementation Notes
|
||||||
|
|
||||||
|
<!-- SECTION:NOTES:BEGIN -->
|
||||||
|
Root cause split in two places: Bun/libsql corrupts large millisecond timestamp strings when coerced through `Number(...)`, and `imm_session_events.ts_ms` being `INTEGER` let runtime event inserts/readbacks return `-2147483648` on CI/runtime.
|
||||||
|
|
||||||
|
Fix shipped by parsing timestamp strings without the broken `Number(largeString)` path, migrating `imm_session_events.ts_ms` to `TEXT`, ordering/retention queries via `CAST(ts_ms AS REAL)`, and avoiding `Number(currentMs)` when reusing already-normalized timestamp strings.
|
||||||
|
|
||||||
|
Added regression coverage for both real runtime event inserts and schema migration/repair of previously truncated session-event rows.
|
||||||
|
<!-- SECTION:NOTES:END -->
|
||||||
|
|
||||||
|
## Final Summary
|
||||||
|
|
||||||
|
<!-- SECTION:FINAL_SUMMARY:BEGIN -->
|
||||||
|
Fixed immersion-tracker timestamp handling under Bun/libsql so large wall-clock millisecond values survive runtime writes, query reads, and schema migration. `bun run test:fast`, `bun run typecheck`, `bun run test:env`, `bun run build`, `bun run test:smoke:dist`, and `bun run changelog:lint` all pass after the patch.
|
||||||
|
<!-- SECTION:FINAL_SUMMARY:END -->
|
||||||
2
vendor/subminer-yomitan
vendored
2
vendor/subminer-yomitan
vendored
Submodule vendor/subminer-yomitan updated: 69620abcbc...ed31b7a3ee
Reference in New Issue
Block a user