Performance regression from seemingly harmless changeset

[michaelwoerister]

#46278 seems to have caused a noticeable performance regression:
http://perf.rust-lang.org/compare.html?start=d5f2745e7ccd247da50f0cb18318ac477e5bccbf&end=5deba220d4c42b5313d7e71731ce5e8698866684&stat=wall-time

It updates a few dependencies (e.g. flate2). Maybe they've become slower?

Anybody know what else this could be about, @rust-lang/compiler?

[alexcrichton]

My money would be on the update to log, it may be the case that 0.3.9 is slower than 0.4.0 because of the shims it introduces (just a hypothesis though). It should be very easy to convert to 0.4.0 though, just bumping cargo.toml requirements throughout the compiler

[MaloJaffre]

I am currently trying to fix this regression.

[andjo403]

was this not fixed with #47161 and #4891 so this can be closed now?

[michaelwoerister]

Yes, it was. Thanks for the hint, @andjo403!