Skip to content
Sign up

flagon-io/g1t

public
00

Where people and agents ship software together. The open-source git platform for the whole job: issues, agents, checks and deploys to the edge.

gitagentscloudflarerustforgemcp
syntaqxAn open pull request's page no longer asks repos whether it is behind a second time, and anyone's ready pull request says it is ready to merge. Signed in, flagon-io/g1t #41 took 5 to 10 seconds and once failed: get_pull worked out whether it is behind (stored with its mergeability for its head, as merge_choices does), then the merge gate asked repos `behind` again on every view, a walk of the git store that took up to 5 s (Workers Logs: /rpc/behind 4.9 s, 649 ms CPU). The gate now takes what the page already worked out (merge_gate_seen), as does the lifecycle's own gate; merging and the queue still ask repos as they did. And the lifecycle was worked out only for pull requests g1t's agent made (managed = 1), which since g1t stopped running agents of its own is none, so a person's pull request that met everything its branch asks for (#41: CI passed, no approvals required) said nothing about being ready. Every open pull request now gets one; only one g1t manages merges by itself (auto_merge is gated on managed), and anyone else's draft still gets none, so its page does not keep refreshing. Lists keep showing a lifecycle only for managed pull requests, since nothing yet refreshes a stored one for the rest. Checked: cargo test -p g1t-work (189 passing).