-
Notifications
You must be signed in to change notification settings - Fork 12.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. Weβll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Performance regression in TS 5.4 #57781
Comments
Would you be able to apply the patch from this PR and test the perf? |
Sure, will try it now |
I applied the patch and this brought typechecking time down to ~331 s. |
Is there any difference there between when you typecheck + emit dts and when you just typecheck? |
Since this is a composite project, subprojects must emit dts for parent projects to use. |
We need a way to repro this in order to investigate. None of our perf tests showed a significant drop in performance over the prior release, so it's something specific to the code here that we'd need to be able to look at to make further progress. |
I can send over any traces or profiles you want so long as I can redact any actual code/code paths. Since I lack the knowledge or tools to identify what the performance hotspots in the code actually are (a frustration that arises frequently when trying to optimize our typechecking perf in general), there's just no way for me to construct a shareable repo that can illustrate the regression. |
If you use https://www.npmjs.com/package/pprof-it on |
Thanks I'll do this over the weekend |
Here are some profiles from pprof-it Before regression After regression |
π Search Terms
performance regression
π Version & Regression Information
β― Playground Link
No response
π» Code
I cannot share our repository nor can I feasibly pare it down to be shareable.
π Actual behavior
Typechecking time in our monorepo takes about 375 seconds on my local machine.
π Expected behavior
Typechecking time used to be about 315 seconds on my local machine with the same code.
Additional information about the issue
Here's my usage of every-ts to bisect the results. It narrows down to #56087. cc @Andarist
The text was updated successfully, but these errors were encountered: