Andy Polyakov
Andy Polyakov
For reference, CUDA failure appears to be transient, it failed to download package index files. They suggest it can be side effect of "Mirror sync in progress".
> How would potential upgrade path for it look like if we choose one version? Not that I can offer any real answers here, but what would be definitely helpful...
> Hm, these failures seem unrelated to my PR? Right, see #677.
> many people have advocated of making the compiler use cc-rs to find the linker If the suggestion is to introduce dependency on cc-rs, then I for one[!] would advocate...
> there already is a dependency, on Windows For reference. The dependency on Windows is of different nature. cc-rs helps rustc to find the toolchain components it's looking for, but...
~CI failures appear to have everything to do with the fact that `cargo test` fails ~in~ on Windows with Rust 1.58, but not with Rust 1.57. I mean it's something...
On a related note one can wonder if it would be appropriate to extend this approach beyond a clang-specific default. Most notably gcc also recognizes `-print-search-dirs` option and cross-gcc-compiler can...
That's unfortunate:-( For reference, the suggestion is a generalization of a problem with wasm target. While `ar(1)` can collect literally any type of files in the .a-archive, it has a...
> If you wanted to make this specific to the wasm target though that seems reasonable to me. Adjusted accordingly. > Making this more targeted that can theoretically get generalized...
In addition I default to llvm-lib in clang-cl compilations. This is an alternative to #668.