Skip to content

fix(todo): retry request timeouts while caller remains active - #235

Merged
tiankaima merged 1 commit into
mainfrom
fix/todo-list-retrieval
Oct 23, 2026
Merged

tiankaima merged 1 commit into
mainfrom
fix/todo-list-retrieval

Conversation

@tiankaima

@tiankaima tiankaima commented Oct 23, 2026 •

Copy link
Copy Markdown
Member

Todo reads stopped after the first HTTP client timeout because it also matches context.DeadlineExceeded. Distinguish the caller's cancelled/expired context from an individual HTTP attempt's timeout, then use the existing three-attempt read budget. Authorization and malformed responses remain non-retryable; mutation delivery is unchanged.

Native HTTP regressions reproduce both header and body timeouts, recover the list-to-numbered-completion flow, stop on caller cancellation/deadline, and prove a timed-out completion stays unknown with one PATCH. CI validation and local checks passed on 8edab66: 1,563 race tests; 19 requirements / 76 native semantic cases; build/vet/lint, deployment/OpenAPI scripts, Compose and generated consistency. Rust fmt/clippy, 46 tests, release build and all 16 current plus seven reference card renders passed.

Part of #233. This fixes a reproduced current timeout-handling defect. The historical reported incident remains unconfirmed, so the Task stays open.

@tiankaima
tiankaima merged commit 415b771 into main Oct 23, 2026
6 checks passed
@tiankaima
tiankaima deleted the fix/todo-list-retrieval branch October 23, 2026 09:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant