Jinjing d4e4f7fc1e fix(gpu): stop replaying the whole crash loop after every update ...

The fallback marker is scoped to appVersion+electronVersion so a new
build gets a fresh hardware attempt. That scoping also discarded what we
knew about the machine: bundle 3GyOWOzGskzdXqlEG8z4UQ shows an update
invalidating the marker, then the full crash burst and relaunch again.
With a ladder that is worse -- each update re-walks it from the bottom.

Record the strongest tier a machine has needed in a version-independent
file. The post-update hardware probe still happens; when it fails the app
resumes at the known tier instead of starting over. Monotonic, so a
shallower later burst cannot talk a known-bad machine back down a rung,
and a hint only -- any read failure degrades to normal escalation.

2026-07-25 12:59:11 -07:00

..

cli

fix(runtime): report the effective ask timeout so a clamped wait isn't misreported (#10550)

2026-07-25 05:42:05 -07:00

main

fix(gpu): stop replaying the whole crash loop after every update

2026-07-25 12:59:11 -07:00

preload

feat(codex): surface a stalled config sync instead of failing silently (#10449)

2026-07-24 18:30:54 -07:00

relay

fix(agent-hooks): install OpenCode status plugin into the WSL guest so status works over WSL (#10328)

2026-07-25 02:30:59 -07:00

renderer

fix(crash-reporting): stop suppressing GPU crashes that the fallback failed to fix

2026-07-25 12:15:08 -07:00

shared

fix(crash-reporting): stop suppressing GPU crashes that the fallback failed to fix

2026-07-25 12:15:08 -07:00

types

…