0 Fork 0

mirror of https://github.com/ksyasuda/SubMiner.git synced 2026-11-10 04:46:07 -08:00

Files

T

sudacode 2a0e931325 refactor(main): drop pass-through main-deps builders (#306 )

* refactor(main): drop mining/anki main-deps builders

* refactor(anilist): drop main-deps builders

* refactor(jellyfin): drop main-deps builders

* refactor(overlay): drop overlay and shortcut main-deps builders

* refactor(main): drop ipc, tray, and osd main-deps builders

* refactor(startup): drop startup main-deps builders

* test(scripts): ignore stray local requests in AUR download test

Also apply prettier to launcher/mpv.test.ts so format:check:src passes.

* fix(shortcuts): forward overlay numeric selection hooks through the composer

composeShortcutRuntimes only passed the digit callbacks, so the visible-overlay
selection added in #78 never ran from the multi-copy/mine-sentence shortcuts.

* fix(stats): skip mpv auto-connect when stats starts the immersion tracker

The immersion-startup main-deps builder never forwarded shouldAutoConnectMpv,
so background stats and stats CLI startup always connected to mpv. Drop the
builder and pass the deps straight to the handler.

* build(deps): bump follow-redirects override to 1.16.2

Resolves GHSA-mrwp-f3q8-3g9p (high), which fails the quality-gate
bun audit --audit-level high step.

* test(subtitles): cover the null-tokenization settle path in prefetch test

The mock ignored the value the test resolved, so the null case never ran.

* test(e2e): surface mpv crashes and install current mpv on macOS CI

The harness IPC client never handled socket close, so a crashed mpv left
commands pending until the 120s test timeout with an empty mpv.log. Reject
pending commands on close, write mpv's own log via --log-file, and refresh
Homebrew on macOS CI so the installed mpv build no longer depends on the
runner image.

* test(e2e): render mpv through cocoa-cb OpenGL on macOS

mpv 0.41.0_12 with libplacebo 7.387.0 segfaults in gpu-next's MoltenVK path
on GitHub's paravirtual macOS GPU on the first frame, so the E2E session
never started.

* test(e2e): name mpv-stdio.log in the mpv close diagnostic

2026-10-31 00:50:33 -07:00

architecture

refactor(main): drop pass-through main-deps builders (#306 )

2026-10-31 00:50:33 -07:00

knowledge-base

test: cut low-value tests, rewrite the rest around behavior, parallelize lanes (#285 )

2026-10-05 00:07:44 -07:00

workflow

refactor(main): drop pass-through main-deps builders (#306 )

2026-10-31 00:50:33 -07:00

README.md

docs: audit and refresh user-facing and internal docs