ci(allocs): allocation tracker measure all transforms#23728
ci(allocs): allocation tracker measure all transforms#23728overlookmotel wants to merge 1 commit into
Conversation
How to use the Graphite Merge QueueAdd either label to this PR to merge it via the merge queue:
You must have a Graphite account in order to use the merge queue. Sign up using this link. An organization admin has enabled the Graphite Merge Queue in this repository. Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue. This stack of pull requests is managed by Graphite. Learn more about stacking. |
570298b to
3f591f9
Compare
There was a problem hiding this comment.
Pull request overview
Updates the memory allocation tracking task to measure allocations for the full set of transformer passes (instead of using an esnext-targeted configuration that skips most transforms), aligning allocation tracking with the intent of detecting allocation regressions/improvements across all transforms.
Changes:
- Switch transformer configuration in the allocation tracker from
TransformOptions::from_target("esnext")toTransformOptions::enable_all(). - Update allocation snapshot outputs for transformer and minifier to reflect the new transformer pass set.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| tasks/track_memory_allocations/src/lib.rs | Uses TransformOptions::enable_all() for transformer runs during warmup and measurement. |
| tasks/track_memory_allocations/allocs_transformer.snap | Updates expected transformer allocation counts under the new configuration. |
| tasks/track_memory_allocations/allocs_minifier.snap | Updates expected minifier allocation counts due to changed pre-minify transformer output/config. |
3f591f9 to
57dfca9
Compare
|
Oh dear. For some reason, this change makes the value for system allocations on |
|
The problem is a |
Allocations tracker was only measuring the allocations in transformer with "esnext" target - i.e. most transforms skipped. Instead, run all transforms, same as we do in benchmarks, so improvements in number of allocations in any transform will register.