You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
find-ui corpus: committed snapshot has drifted from upstream #807
No action is required. The corpus is refreshed automatically by scripts/build-cli.ps1 on the next -Stable release, and users with network
access already get live data — the baked snapshot is only the offline floor.
This issue exists so a large or unexpected swing (a source losing most of its
scenarios, for example) is visible before it ships.
Closing: this was working as designed, and the design was wrong.
Upstream gaining samples is the steady state between releases, not an event. The corpus is re-baked by scripts/build-cli.ps1 on every -Stable release and users with network access already get live data, so this issue was open essentially permanently and collecting a weekly "still present" comment that no one could act on — on the same find-ui-corpus label that carries the genuine alerts.
The drift job now reports routine drift to the workflow run summary only. It still opens an issue for the two actionable cases:
the bake fetched nothing (unchanged), and
a source lost more than 25% of its scenarios, or vanished from an otherwise successful bake.
That second case is worth keeping: SnapshotBaker.BakeAsync only fails a source that returns zero scenarios, so a fetcher that survives an upstream restructure while matching a fraction of what it used to bakes green today and would ship a gutted source.
Activity
github-actions commentedon Sep 7, 2026
Still present as of the latest scheduled check.
snapshot-gallery.json.br— 321 → 323 scenarios (+2)snapshot-reactor.json.br— 93 → 95 scenarios (+2)Workflow run
Jaylyn-Barbee commentedon Sep 8, 2026
Closing: this was working as designed, and the design was wrong.
Upstream gaining samples is the steady state between releases, not an event. The corpus is re-baked by
scripts/build-cli.ps1on every-Stablerelease and users with network access already get live data, so this issue was open essentially permanently and collecting a weekly "still present" comment that no one could act on — on the samefind-ui-corpuslabel that carries the genuine alerts.The drift job now reports routine drift to the workflow run summary only. It still opens an issue for the two actionable cases:
That second case is worth keeping:
SnapshotBaker.BakeAsynconly fails a source that returns zero scenarios, so a fetcher that survives an upstream restructure while matching a fraction of what it used to bakes green today and would ship a gutted source.The +2/+2 drift reported here needs no action.
Stop filing issues for routine find-ui corpus drift (#825)