Skip to content

test(combo): update the failover e2e assertion the effort fix invalidated - #3175

Merged
lidge-jun merged 1 commit into
devfrom
codex/3108-followup-e2e-assertion
Sep 1, 2026
Merged

lidge-jun merged 1 commit into
devfrom
codex/3108-followup-e2e-assertion

Conversation

@lidge-jun

Copy link
Copy Markdown
Owner

Summary

CI on dev caught one assertion #3172 should have updated and did not: tests/server-combo-failover-e2e.test.ts > fresh child reparsing recomputes vision and effort per target expected the first target to receive no reasoning_effort.

That expectation encoded the defect #3172 fixed. In that fixture the combo's members advertise a: ["low"] and b: ["low", "high"], so the catalog intersects them to ["low"] and effectiveComboDefault("high", ["low"]) resolves to low — the served catalog advertises low for this combo. The old request path tested literal membership, missed, and sent nothing, so the provider default applied. Sending low to the first target is what the catalog promised all along.

The assertion is updated to "low" with a comment recording why, rather than the fix being reverted. The second target still receives high, which is what that half of the test was really checking.

This is the follow-up to #3172 (7386b52). No source change.

Verification

Exact head:

  • bun test tests/server-combo-failover-e2e.test.ts — 74 pass, 0 fail, 453 expect() calls. The failing case reproduced locally before the change and passes after.
  • bun test tests/combos.test.ts tests/combo-stream-preflight.test.ts tests/combo-child-headers.test.ts — 51 pass, 0 fail.

Found by CI on the merged dev head rather than by local review — the scoped local runs for #3172 covered tests/combos.test.ts and the catalog and boundary suites, but not this end-to-end failover file.

Checklist

  • Targets dev
  • Test-only change; no source, config, schema, or dependency change
  • The corrected expectation is explained in the test itself, so the next reader does not re-assert the old behavior
  • No credential, auth, workflow, or release-automation surface touched

…ated

The case expected the first target to receive no reasoning_effort, which is the
behavior #3172 removed. That combo's members advertise low and low+high, so the
catalog advertises low, and low is now what the first target receives.

Refs #3108
@lidge-jun
lidge-jun requested a review from Ingwannu as a code owner September 1, 2026 16:41
@lidge-jun
lidge-jun merged commit 22a643a into dev Sep 1, 2026
1 check passed
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Sep 1, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-09-01T16:44:44.718118Z a18dfd2 PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@lidge-jun
lidge-jun deleted the codex/3108-followup-e2e-assertion branch September 1, 2026 16:41
@github-actions

github-actions Bot commented Sep 1, 2026

Copy link
Copy Markdown
Contributor

✅ Deterministic PR hygiene checks passed.

@github-actions github-actions Bot added the chore Maintenance, CI, tests, refactors, or build changes (not a user-facing bug or feature). label Sep 1, 2026
tarunravi pushed a commit to tarunravi/opencodex that referenced this pull request Sep 14, 2026
…ated (lidge-jun#3175)

The case expected the first target to receive no reasoning_effort, which is the
behavior lidge-jun#3172 removed. That combo's members advertise low and low+high, so the
catalog advertises low, and low is now what the first target receives.

Refs lidge-jun#3108

Co-authored-by: jun <jun@lidge.dev>
agentHits pushed a commit to agentHits/opencodex that referenced this pull request Sep 17, 2026
…ated (lidge-jun#3175)

The case expected the first target to receive no reasoning_effort, which is the
behavior lidge-jun#3172 removed. That combo's members advertise low and low+high, so the
catalog advertises low, and low is now what the first target receives.

Refs lidge-jun#3108

Co-authored-by: jun <jun@lidge.dev>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

chore Maintenance, CI, tests, refactors, or build changes (not a user-facing bug or feature).

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant