Repository navigation
perf(ci): Split large Node integration test suites - #26420
Conversation
Co-Authored-By: OpenAI Codex <codex@openai.com>
size-limit report 📦Compared: e78d40c (PR test merge using develop at 1d3a542) Against: 1d3a542 (develop) · base run #66400 ✅ Baseline matches: this diff contains only this PR's changes.
|
Co-Authored-By: OpenAI Codex <codex@openai.com>
Optional E2E testsSome optional checks were not successful. They do not block the merge. 1 not successful, 62 successful. See the run.
Successful checks
|
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
There are 3 total unresolved issues (including 2 from previous reviews).
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 9252e89. Configure here.
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
809a3be to
edd5fe4
Compare
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
Co-Authored-By: OpenAI Codex <codex@openai.com>
JPeer264
left a comment
There was a problem hiding this comment.
Overall LGTM. I'd still prefer to have the assertions directly in the tests, rather than having exports and then use them in the tests.
| } | ||
|
|
||
| describe('express tracing', () => { | ||
| export function registerExpressRoutingTests(): void { |
There was a problem hiding this comment.
l: Is there any benefit of having that in named exports (other than keeping the LOC changes low)? I'd prefer to have them actually directly in the tests. IMO having request-data/test.ts and all its assertions is fine.
There was a problem hiding this comment.
for vercel AI it makes sense to have it like this because v6 and v7 share all the assertions. other than that no so I'll restructure the remaining ones
There was a problem hiding this comment.
done, now look at that mighty diff
Co-Authored-By: OpenAI Codex <codex@openai.com>

What
Split long Node integration suites into smaller files so the existing shards can distribute work more evenly. As an example, before we had single test files for vercel ai that took 2-3 minutes to execute. Playwright shards split by test files so this makes the load distribution very uneven. We should prefer smaller files.
Why
The slowest required Node integration test step fell 184→162s (−22s, 12%); the longest complete Node integration job fell 229→198s (−31s).
Baseline → final benchmark. Single-run measurements; test-step times exclude setup/queueing, complete-job times include setup/cleanup. Inventories and pass/skip states match.