Repository navigation
Revert "Report CPU limits hit in a microtask at startup and dynamic import (#8216)" - #8488
Conversation
There was a problem hiding this comment.
💬 Approved with comments — Clean, intentional revert of #8216. The only note is informational: in the restored main-module startup code, the limit check runs before the limit is recorded, so it never fires, and CPU limits hit in microtasks at startup and during dynamic import go unreported again, as they did before #8216.
Findings: 1 info (1 is an inline comment on the diff)
The decision is advisory: this review is a comment and never approves, blocks or requests changes on the pull request.
📋 Code Quality: 💬 1 info
Exact revert; restored main-module limit check is effectively inert
Recommendation: 💬 Optional Nits
- [INFO] Restored KJ_DEFER is declared after limitScope, so it runs before the limit error is recorded and won't catch microtask limit hits at startup (
src/workerd/io/worker.c++:2303)
View full review
This reverts #8216 exactly and brings back the earlier behaviour. CPU-limit errors raised while microtasks drain are again not reported:
- on non-module startup (no check after the limit scope closes);
- during dynamic import (no check after
handler()); - at main-module startup, where the restored
KJ_DEFERis declared afterlimitScope.
That last point is noted inline. If the revert is intended, none of this is a defect, but a short rationale in the description would help.
✅ All other checks passed
📝 Documentation: ✅ LGTM
No documentation concerns.
Reviewed commit: ac43e081
This reverts commit dcc41ac.
Paired with the edgeworker branch of the same name, which drops the patch that currently carries this revert.
🤖 Generated with Claude Code