| 26 Feb 2026 |
 emily | I remember there's something awful like that and it causes headaches | 20:18:15 |
| emily | I think Artturin exposed me to that particular horror | 20:18:24 |
 @rosssmyth:matrix.org | Yes, rustup ships a full mingw toolchain in the distribution bundle (including libpthread) | 20:18:25 |
| emily | (why does Rust even use libpthread on MinGW if it can use native threads on the MSVC target; is it required to use their shim layer for interoperability or something…) | 20:18:55 |
| @rosssmyth:matrix.org | It is required because of mingw | 20:19:07 |
| @rosssmyth:matrix.org | Basically most application have no clue what msvc is | 20:19:19 |
| @rosssmyth:matrix.org | * Basically most applications have no clue what msvc is | 20:19:31 |
| emily | yeah but most Rust applications are like, written in Rust | 20:19:32 |
| emily | and even most C libraries that Rust applications link to probably know what an MSVC is | 20:19:38 |
| @rosssmyth:matrix.org | Sure, but if you want to interoperate with other mingw libs written in C you must play by their rules | 20:19:57 |
| emily | I don't like how MinGW can't decide if it wants to be a free toolchain for Windows or half a Cygwin. eh whatever | 20:20:07 |
| @rosssmyth:matrix.org | Same way as using musl and glibs | 20:20:13 |
| emily | anyway I think someone should ask rustc what linker command it's actually calling | 20:20:23 |
| @rosssmyth:matrix.org | * Same way as using musl and glibc | 20:20:27 |
| @rosssmyth:matrix.org | It's probably in our rustc wrapper somewhere I'd guess | 20:20:44 |
| @rosssmyth:matrix.org | I'll poke around | 20:20:47 |
| emily | isn't there just a flag that makes it print the command it runs? | 20:21:28 |
| emily | I thought it did that whenever it failed even | 20:21:34 |
| @rosssmyth:matrix.org | Yeah | 20:21:37 |
| @rosssmyth:matrix.org | rustc -V or something | 20:21:44 |
| @rosssmyth:matrix.org | rustc --print link-args but you need to supply it with some files to munch on | 20:24:41 |
 dramforever | i can't compile ripgrep rn because it depends on ... arm64ec wine when crossing from aarch64-linux? which we don't have | 20:26:20 |
| dramforever | but pkgsCross.mingwW64.bonk works for me | 20:26:29 |
| dramforever | * i can't compile ripgrep rn because it depends on ... arm64ec wine plus an x86_64 emulator when crossing from aarch64-linux? which we don't have | 20:26:47 |
| dramforever | did this magically get fixed in the few months...? | 20:27:20 |
| dramforever | * did this magically get fixed in the few months since that PR...? | 20:27:26 |
| emily | iirc we have some awful thing where you can pick between five different thread implementations for MinGW? | 20:27:35 |
| emily | so maybe that's relevant | 20:27:38 |
| @rosssmyth:matrix.org | I removed most of those, so it should only use the real one most of the time except in one weird case | 20:28:01 |
| @rosssmyth:matrix.org | Which idk why it is needed, but I just didn't touch it. | 20:28:21 |
