Rust 1.50.0 pre-release testing

You can find the announcement on the Inside Rust blog:

7 Likes

FWIW, the linked index.html appears empty -- but I was still able to manually download the source package rustc-1.50.0-src.tar.xz anyway.

That's weird, I can see the content if I open the index on my browser. Is anything weird going on in the network tab of the browser's developer tools?

I am reproducibly seeing the following failure from x.py build against the 1.50.0 source dist in my work environment (where we've built just about every release since 1.11.0 successfully). It fails at rust/lib.rs at 690501e3e9535389a1441cee74d049ba16b963fa · rust-lang/rust · GitHub after "Assembling stage1 compiler". Still investigating but I figured I'd raise it in case anyone is seeing the same.

Assembling stage1 compiler (x86_64-unknown-linux-gnu)
thread 'main' panicked at 'src.symlink_metadata() failed with No such file or directory (os error 2)', src/bootstrap/lib.rs:1192:24
stack backtrace:
   0: rust_begin_unwind
             at /rustc/e1884a8e3c3e813aada8254edfa120e85bf5ffca/library/std/src/panicking.rs:495:5
   1: std::panicking::begin_panic_fmt
             at /rustc/e1884a8e3c3e813aada8254edfa120e85bf5ffca/library/std/src/panicking.rs:437:5
   2: bootstrap::Build::copy
             at ./rust/src/bootstrap/lib.rs:1192:24
   3: <bootstrap::compile::Assemble as bootstrap::builder::Step>::run
             at ./rust/src/bootstrap/compile.rs:1025:13
   4: bootstrap::builder::Builder::ensure
             at ./rust/src/bootstrap/builder.rs:1488:23
   5: bootstrap::builder::Builder::compiler
             at ./rust/src/bootstrap/builder.rs:579:9
   6: <bootstrap::compile::Std as bootstrap::builder::Step>::make_run
             at ./rust/src/bootstrap/compile.rs:48:23
   7: bootstrap::builder::StepDescription::maybe_run
             at ./rust/src/bootstrap/builder.rs:179:13
   8: bootstrap::builder::StepDescription::run
             at ./rust/src/bootstrap/builder.rs:200:25
   9: bootstrap::builder::Builder::run_step_descriptions
             at ./rust/src/bootstrap/builder.rs:571:9
  10: bootstrap::builder::Builder::execute_cli
             at ./rust/src/bootstrap/builder.rs:562:9
  11: bootstrap::Build::build
             at ./rust/src/bootstrap/lib.rs:510:13
  12: bootstrap::main
             at ./rust/src/bootstrap/bin/main.rs:30:5

Full output: https://paste.rs/qzm

Looks like not a problem — this is from rust-lang/rust#77117 and similar to rust-lang/rust#80086. We just needed to provide an llvm-dwp binary in the right place. It would be convenient if llvm-dwp were optional.

Both in Chrome and Firefox, I get only the headers "Name", "Size", "Last modified", and the separator line below that, with no actual files listed. Nothing looks weird in the network tools, just a 404 for favicon.ico. For comparison, the index for 1.49 at 2020-12-29 still looks fine.

Following up on llvm-dwp — the way that llvm-dwp is located by bootstrap in this release seems wrong. I filed rust-lang/rust#81949 to follow up.

@cuviper: I see the same empty page with headers as you do. I live "nearby". Hypothesizing that some datacenter/whatever infrastructure in our part of the world somehow cached that index page prior to the release going up, and Pietro is hitting a different one...?

1 Like

index.html works ok for me, here on the east coast. I suppose we can issue an invalidation just in case but I'm not too worried about it.

Opened a PR to rebuild the stable artifacts with the fix for the bug reported by @dtolnay. Hopefully the second pre-release will be out today.

I just noticed something missing from the 1.50 release notes: they only mention IpAddr::is_ipv4 and is_ipv6 being made const, while in fact all (stable) methods of IpAddr, Ipv4Addr and Ipv6Addr were made const.

Is this something that should be updated? If so, how should that be done.

You can open a Pull Request against rust/RELEASES.md at master · rust-lang/rust · GitHub adding the missing entries. It looks like PR #79342 maybe somehow slipped through.