Skip to content

Isolate HTTP notes and bound Git transport - #2116

Open
svarlamov wants to merge 5 commits into
mainfrom
feat/git-notes-http-isolation-timeouts
Open

Isolate HTTP notes and bound Git transport#2116
svarlamov wants to merge 5 commits into
mainfrom
feat/git-notes-http-isolation-timeouts

Conversation

@svarlamov

@svarlamov svarlamov commented Aug 10, 2026

Copy link
Copy Markdown
Member

Summary

  • Route HTTP-backed reads, searches, traversal, fetches, and CI synchronization exclusively through SQLite and HTTP APIs.
  • Remove Git-note fallbacks and display-ref materialization; git-ai notes migrate remains the sole explicit migration exception.
  • Add hard timeouts and process-group cleanup for Git-notes fetch/push transport.
  • Add regression coverage proving HTTP-mode commands never mention Git-notes refs.

Verification

  • task build
  • task test TEST_FILTER=http_backend
  • task lint

Stack 1/2; followed by #2117.

Part of #2099.

Route HTTP-backed note reads, searches, traversal, and CI synchronization exclusively through SQLite and HTTP APIs. Remove Git-note fallbacks and display-ref materialization, while retaining git-ai notes migrate as the explicit migration exception.\n\nAdd timeout-aware Git transport execution and regression coverage proving HTTP log paths never spawn commands that mention Git-notes refs.\n\nPart of #2099.
devin-ai-integration[bot]

This comment was marked as resolved.

Cache notes-backend dispatch until its config inputs change so repeated note lookups do not rebuild and recompile the full runtime config. Preserve immediate backend switching through the cache key.\n\nGive timed internal transports a null stdin, matching Command::output behavior and preventing detached children from stalling on terminal credential prompts.
devin-ai-integration[bot]

This comment was marked as resolved.

Apply the standard non-interactive Windows creation flags to timed internal commands so background notes transport cannot flash a console window.\n\nMake the shared authorship fetch entry point warm the HTTP cache and report success, including for internal machine-command callers. Add an end-to-end cache-warming regression.
devin-ai-integration[bot]

This comment was marked as resolved.

devin-ai-integration[bot]

This comment was marked as resolved.

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