Daniel Silverstone

We set `RUSTC` in our proxies to improve performance because it means `cargo` doesn't have to indirect through Rustup again for every compilation step. My gut feeling, if you are...

We use `/proc/self/exe` because that tends to tell us the **userland** host type rather than `uname -m` which would tell us the kernel architecture. Yes it's possible that's correct, but...

Perhaps we could switch from reading `/proc/self/exe` to reading `$SHELL` - would there be any situations we can think of where that wouldn't work?

I have been thinking about custom toolchain origins, and also about aliasing and relabelling toolchains in general. I don't have a good answer to all this yet; the start of...

How did the `cargo` part of this go Eric? I'm reluctant to start validating this until we know Cargo makes sense.

There's no guarantee that the linker you specified is (a) correct or (b) available on the user's system. Sadly cross-compilation like this is not easy to make reliable, and I...

While this is definitely irritating for you; it's not a bug in Rustup. It should not be possible for a userland program inside a VM inside a modern OS to...

For now I'm going to pin this issue so that people notice it (hopefully), but otherwise there's nothing to be done I think

Closed - we can revisit if/when tls1.3 is sufficiently widespread that we're not risking hurting people

If you are doing cross-builds from within invocations of `cargo` are you perhaps not being sure to clean your environment of things like `RUSTC` `RUSTDOC` `RUSTFLAGS` or other environment variables...