kevinjqliu commented on PR #3520:
URL: https://github.com/apache/iceberg-rust/pull/3520#issuecomment-6446002347
Checking some results.
This PR was merged at 18:33Z yesterday. I compared about 1,600 rust-cache
restores from CI, Public API and Bindings Python runs on `main`, PRs and the
merge queue, from Oct 22 onward.
- **One cache key per job and OS.** Before the merge, every job/OS pair had
two keys, one for each stable version on the runners (1.99.0 and 1.99.1). Now
each has one. On macOS, two different runner image versions ran after the merge
and both used the same key.
- **Misses are gone.** Once `main` had saved the new caches (by about
19:30Z), none of the 191 restores missed.
| | full hit | partial hit | miss |
| --- | --- | --- | --- |
| Before | 64% | 18% | 18% |
| Warm-up (first hour) | 16% | 13% | 71% |
| After warm-up | 89% | 11% | **0%** |
The jobs that missed most now always hit, and their average time dropped:
| Job | Misses before | Avg time before → after |
| --- | --- | --- |
| `check-public-api` | 44% | 3.3 → 1.5 min |
| Bindings Python, macOS | 46% | 3.0 → 2.5 min |
| `build_with_no_default_features`, macOS | 34% | 1.3 → 0.8 min |
`Tests (default)`, `Tests (doc)` and `check_standalone` each also got about
0.5–0.7 min faster on average.
😄
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
