Skip to content

Gengchou v2.2.2

Choose a tag to compare

@github-actions github-actions released this 17 Jul 05:14
Immutable release. Only release title and notes can be modified.

Gengchou v2.2.2

This patch makes multi-provider refreshes complete sooner and keeps Claude
quota data aligned with the provider's real cooldown and reset boundaries.

Fixes

  • Poll enabled providers concurrently so one slow endpoint no longer delays
    fresh data from every other provider in the same refresh pass.
  • Expire a cached Claude quota snapshot as soon as one of its reported reset
    times passes instead of retaining pre-reset usage until the next cooldown.
  • Align the one-minute refresh timer with Claude's 180-second normal and
    120-second fast cooldown deadlines, avoiding an extra scheduler interval
    when a deadline falls between fixed ticks.
  • Preserve existing partial-success behavior, provider-specific error badges,
    and Claude rate-limit backoff while applying the new scheduling logic.
  • Retry one transient Codex TLS, transport, or 502/503/504 failure after a
    short delay, with a Codex-specific 10-second request timeout so a stalled
    usage endpoint cannot hold the refresh pass for the previous 30 seconds.
  • Keep the last known number visible with a subtle ~ marker for transient
    provider failures and rate limiting, regardless of outage duration or cache
    availability. Reserve the red ! for states that require user action, such
    as missing, expired, or rejected credentials.
  • Require consistent authentication rejection across Antigravity's fallback
    endpoints, and avoid applying one provider's authentication error to every
    provider when an all-provider refresh fails for mixed reasons.
  • Reuse Antigravity connections with a 10-second per-request timeout, and
    suppress repeated Codex or Antigravity network calls for credentials already
    rejected by the provider. A credential change retries on the next refresh;
    partial failures with unchanged credentials receive a 15-minute safety
    recheck while the other providers continue refreshing.
  • Log consecutive transient failures and recovery transitions so short network
    blips can be distinguished from persistent provider outages.
  • Detect when the secondary monitor hosting the taskbar widget, floating
    monitor, or an open detail popup is removed (including RDP topology changes)
    and reset the affected surface to its primary-screen default. Ordinary DPI
    changes and Explorer taskbar recreation keep the user's chosen position.

No settings or cache migration is required. Existing v2.2.1 installations can
update in place through Gengchou's built-in updater.