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. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

All langs: apply the explicit-this patch to all remaining code #7245

Open

erik-krogh wants to merge 2 commits into github:main from erik-krogh:explicit-this-all-the-places

+478 −450

Contributor

erik-krogh commented Nov 26, 2021

The patch was written automatically by a tool.


          add all remaining explicit this

6ff8d4d

github-actions bot added C# C++ Java JS Python Ruby labels

Contributor

owen-mc commented Nov 26, 2021

❤️


          revert explicit-this that caused non-monotonic recursion

74158f1

erik-krogh marked this pull request as ready for review

erik-krogh requested review from as code owners

erik-krogh added the no-change-note-required label

Contributor

esbena commented Nov 29, 2021

JS 👍🏻

esbena reviewed

java/ql/lib/semmle/code/java/security/CommandArguments.qll Outdated Show resolved Hide resolved

hvitved approved these changes

Contributor

hvitved left a comment

C# 👍 Could you please start DCA runs for all languages as well? We have seen in the past that adding explicit this qualifiers made performance worse.

MathiasVP approved these changes

Contributor

MathiasVP left a comment

C++ 👍 (assuming DCA is happy)

Contributor

aschackmull commented Nov 30, 2021

I think the dca run ought to have included more queries for Java to test more broadly as these changes are all over. As-is there are 3 projects that show minor slowdowns, which may be significant, but I'm not entirely sure whether I ought to consider that suspicious or not. A 10% slowdown with the absolute slowdown being twice the stddev of 3 runs seems like it might be significant. The two 25% slowdowns could have benefitted from a couple of reruns to get stddev values - as-is it's hard to say whether to read something into those.

Contributor Author

erik-krogh commented Nov 30, 2021

I think the dca run ought to have included more queries for Java to test more broadly as these changes are all over.

I've queued some retries to get some stddev values on those regressions.

Evaluations for the other languages look OK (some improvements, some regressions, seems to be within margin of error).

Except for csharp, which has some massive performance improvements.
But those improvements are spurious and are caused by the second database missing a bunch of source-code (see the source-archive-hashes folder in the DCA report).
I've started a new DCA experiment where the variants are not grouped.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment