Prepare beta 1.51.0

[pietroalbini]

cc @rust-lang/release
r? @ghost

[pietroalbini]

@bors r+ p=1000 rollup=never

[bors]

📌 Commit bae1a0a has been approved by pietroalbini

[pietroalbini]

@bors r-

[pietroalbini]

@bors r+ p=1000 rollup=never

[bors]

📌 Commit 4745585 has been approved by pietroalbini

[bors]

⌛ Testing commit 4745585 with merge 1dddb472b0e764de5d2029560576f36a5b706cb9...

[pietroalbini]

Also needed to cherry-pick 483c1a8 to make the build work. Will send a PR landing that commit on master in a bit.

@bors r+ p=1000 rollup=never

[bors]

📌 Commit e708e2a has been approved by pietroalbini

[bors]

⌛ Testing commit e708e2a with merge 30b51f4208197b622efafc707a28c08007f2cb37...

[bors]

💔 Test failed - checks-actions

[pietroalbini]

Seems like rustfmt depends on an old version of rustc-ap-rustc_arena, which used the now-removed array_value_iter feature.

[Mark-Simulacrum]

Hm. cc @calebcartwright

It looks like we'll probably want to either bump rustfmt or do some other fix, not really sure...

[ollie27]

Did you intend to apply this PR to the master branch?

[bors]

⚠️ The base branch changed to beta, and the PR will need to be re-approved.

[Xanewok]

#81768 has been merged and RLS/Rustfmt toolstate is green now.

[pietroalbini]

Thanks! Cherry-picked the following two PRs:

• update RLS and rustfmt update RLS and rustfmt #81768 (fixes the build)
• bootstrap: Locate llvm-dwp based on llvm-config bindir bootstrap: Locate llvm-dwp based on llvm-config bindir #81955 (beta-accepted)

@bors r+ p=500 rollup=never

[bors]

📌 Commit 9e98dfe has been approved by pietroalbini

[pietroalbini]

@bors r+ p=500 rollup=never

[bors]

📌 Commit e15d61c has been approved by pietroalbini

[bors]

⌛ Testing commit e15d61c with merge db5d098b874fab52e044fca0b96644bef0503c3a...

[bors]

💔 Test failed - checks-actions

[Xanewok]

This is an instance of #79132 (comment), which probably needs a 7477867 backport for now.

I'm sorry that RLS doesn't build on aarch64-pc-windows-msvc proper - this was somehow alleviated by rust-lang/rls#1713 but not completely fixed - we ideally would upgrade to Tokio 1.0 (along with core jsonrpc-* deps) but that's not possible due to parity-tokio-ipc being blocked on paritytech/parity-tokio-ipc#30 (which is effectively blocked on Tokio design work in tokio-rs/tokio#3118).

However, I believe the RLS should build (but not pass tests, as these unconditionally require Tokio) with --no-default-features: this disables (only) ipc feature, which allows RLS to spawn rustc as child processes rather than executing them all in-process. This offers considerable benefits when checking wide (package-wise) workspaces but ultimately is an optimization and should work correctly without it.

We might:

1. Try distributing RLS with --no-default-features on aarch64-pc-windows-msvc for the time being (without blocking the CI on RLS tests)
2. Backport the hack from 7477867 and wait for the Tokio named pipes design work to conclude in the meantime
3. Quickly write a hacky blocking IPC (sans Tokio) solution for named pipes, which I unfortunately don't have time for in the next couple of days

Solution no. 2 seems easiest and is proven to work (#79132) but if we try going with solution no. 1, if we don't need to hack dist.rs much, then we might end up with working RLS for the aarch64-pc-windows-msvc target until we fix the ipc feature there.

[pietroalbini]

Backported 7477867.

@bors r+ p=500 rollup=never

[bors]

📌 Commit b2d3220 has been approved by pietroalbini

[bors]

⌛ Testing commit b2d3220 with merge a5a775e...

[bors]

☀️ Test successful - checks-actions
Approved by: pietroalbini
Pushing a5a775e to beta...