search

rust-lang / rust

Empowering everyone to build reliable and efficient software.
https://www.rust-lang.org
Other
98.61k stars 12.74k forks source link

regression: std no longer in toolchain dir #132202

Open Mark-Simulacrum opened 3 weeks ago

Mark-Simulacrum commented 3 weeks ago

https://crater-reports.s3.amazonaws.com/beta-1.83-3/beta-2024-10-19/reg/bevy_dexterous_developer_library-0.4.0-alpha.3/log.txt

root is in prefer-dynamic crate, probably just needs a compat note

apiraino commented 2 weeks ago

I quickly skimmed at the most recent merges with relnotes and didn't find anything from {t-libs,-api} about that. Very likely have missed that, just wanted to check if it's tagged for the changelog

jieyouxu commented 2 weeks ago

I quickly skimmed at the most recent merges with relnotes and didn't find anything from {t-libs,-api} about that. Very likely have missed that, just wanted to check if it's tagged for the changelog

As discussed in https://rust-lang.zulipchat.com/#narrow/channel/245100-t-compiler.2Fwg-prioritization.2Falerts/topic/.23132202.20regression.3A.20std.20no.20longer.20in.20toolchain.20dir, this is probably T-bootstrap (and maybe weakly T-compiler) change from #131188. But yes, we should probably add a compat note.

cc @Kobzol as you may know more about this, but I suspect the prefer-dynamic crate was relying on undocumented unguaranteed behavior and that this isn't considered a "breakage" in the strict sense.

EDIT: nevermind @bjorn3 already commented on the prefer-dynamic crate https://github.com/WilliamVenner/prefer-dynamic/issues/2, oops.

Kobzol commented 2 weeks ago

Yeah I think that if anyone depended on libstd.so within the lib directory, it was an instance of Hyrum's Law.