Repository navigation
chore: merge main into v3 - #1135
Conversation
Co-authored-by: evlogai[bot] <313167257+evlogai[bot]@users.noreply.github.com>
…v3-beta3d Amp-Thread-ID: https://ampcode.com/threads/T-01a1ede8-9e2e-73d6-b5bf-f813b0237eb2 # Conflicts: # packages/evlog/test/helpers/timers.test.ts # packages/evlog/test/helpers/timers.ts
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the
|
Risk score: Very high (4/4)Assessed at 7f8e5d1 from the changed files and the title, by
|
|
Thank you for following the naming conventions! 🙏 |
@evlog/cli
@evlog/eve
evlog
@evlog/nuxthub
@evlog/signals
@evlog/telemetry
commit: |
Brings the three commits on
mainthatv3doesn't have intov3before beta.3 (#1017): #1132 (content-lint keeps underscores in heading anchors), #1122 (re-measure a linear-time check before failing it on one noisy sample) and #1133 (drop an em dash from Evi's docs connection description). Like #1121 and #1129, this lands by fast-forwardingv3to this branch's head, not by a squash, somainstays an ancestor ofv3.None of the three touches a published package or has a changeset, so beta.3's versions and changelog stay the same.
Conflict
packages/evlog/test/helpers/timers.tsandtimers.test.ts: #1122 onmainand #1118 onv3both changedexpectLinearTime. The code merged cleanly to both behaviours:v3's 32x ratio (#1118, for Node 22's slower long strings) andmain's up to three measurements (#1122). Only the doc comment and two neighbouring tests conflicted:main's "measures again" test the slow first sample is now 5x instead of 4x. At 4x the first ratio is exactly 32 (8 × 4), which missesv3's limit only because the check is strict<. At 5x (ratio 40) it misses clearly, so the test still proves the retry.Validation
At this branch's head:
With
MAX_ATTEMPTSset to 1, the "measures again" test fails (1 failed | 11 passed), so it still checks the retry. CI runs the rest.ELI5
Three small changes landed on the 2.x branch, so this copies them into the 3.0 branch before the next beta. Both branches had changed the same timing test helper in different ways, and this keeps both changes.