rusackas commented on PR #45606: URL: https://github.com/apache/superset/pull/45606#issuecomment-6332350254
`lint-frontend` numbers from this PR's run (03:59 UTC), against the master run at 00:20 UTC when runners were quick and against the three sibling PRs that ran between 03:10 and 03:45 UTC when they were not: | | master, 00:20 | sibling PRs, 03:10 to 03:45 | this PR, 03:59 | |---|---|---|---| | `lint` step (`npm i` + oxlint + custom rules) | 19 s | ~5 min | 7 s | | `tsc` step (`npm i` + plugins:build + tsc) | 154 s | 409 to 459 s | 155 s | | `lint-frontend` job | 3m34s | 12m47s to 13m39s | 4m13s | So the `npm i` costs anywhere from a dozen seconds to five minutes per step depending on how the registry and the runner are doing, and this PR removes it from both steps outright. The 155 s left in the `tsc` step is babel (28 s), `tsc --build` (68 s) and `tsc --noEmit` (58 s); #45603 takes the two tsc phases down to 30 s and 18 s. The one cost on this side is the image build when the lockfile or a workspace manifest changes, since the `npm ci` layer is keyed on those: 172 s here versus about 50 s on a cached layer. That's per change, not per run, and it's the same layer that already rebuilt on every dependency bump. -- 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]
