Bring back x86_64-sun-solaris target to rustup

[kulikjak]

Change #82216 removed now deprecated target x86_64-sun-solaris from CI, thus making it no longer possible to use $ rustup target add x86_64-sun-solaris to install given target (see #85098 for details). Since there should be a period of time between the deprecation and removal, this PR brings it back (while keeping the new one as well).

Please, correct me if I am wrong; my assumption that these Docker scripts are being used to build artifacts later used by rustup might be incorrect.

Closes#85098.

[rust-highfive]

r? @kennytm

(rust-highfive has picked a reviewer for you, use r? to override)

[kennytm]

r? @nagisa

[nagisa]

r? @Mark-Simulacrum

[Mark-Simulacrum]

Preemptively nominating for beta, but will need to review later.

[jhpratt]

If this is just an issue with the binary being available, this technically wouldn't need to be tied to a release, correct? I think having a build for Rust 1.52 would be useful, as otherwise there would be a seemingly arbitrary gap in the availability of the target.

[Mark-Simulacrum]

I don't think it's likely that we'll produce another point release this cycle, so there probably won't be 1.52 artifacts published. cc @rust-lang/release for thoughts on that.

[Mark-Simulacrum]

This PR looks OK to me otherwise, though I note it doesn't attempt to alter the build toolchain for the illumos target back to sun from pc, which changed in https://github.com/rust-lang/rust/pull/82216/files#diff-f7fe7525804df85adb4a03efa70e96ee893009d6260066cde665abb8088b8cbeR21.

I admit to not really knowing if the build toolchain makes any difference - I don't think that PR changed anything else. Certainly we don't ship -pc-illumos and -sun-illumos or anything like that...

cc @steveklabnik @jculow @pmooney

[jhpratt]

Perhaps I should be a bit clearer on what I was saying. Why does this need a point release at all? All this does is generate the binary that rustup then downloads, correct? If so, what's to prevent the creation of a binary for Rust 1.52.0/1.52.1? Checksums for something?

[Mark-Simulacrum]

We don't have the ability right now to partially edit an existing release, not sure whether it's completely impossible due to checksums, but not something we can do right now on an implementation level.

[Mark-Simulacrum]

@bors r+ rollup=never

Alright, let's get this merged in. I think it's an improvement -- not quite back to where we where before, particularly on the illumos target, but having not heard back there I'm inclined to let that stay as-is. It's not clear how much the toolchain that builds the compiler matters for those purposes, and regardless, further work is always possible in future PRs -- for now this should get the breakage in 1.52 due to a missing tier-2 target fixed.

Approving for beta backport so we only have one 'missing' release -- 1.52.

[bors]

📌 Commit b13185d has been approved by Mark-Simulacrum

[bors]

⌛ Testing commit b13185d with merge 54bdfa1...

[bors]

☀️ Test successful - checks-actions
Approved by: Mark-Simulacrum
Pushing 54bdfa1 to master...

[kulikjak]

This PR looks OK to me otherwise, though I note it doesn't attempt to alter the build toolchain for the illumos target back to sun from pc, which changed in https://github.com/rust-lang/rust/pull/82216/files#diff-f7fe7525804df85adb4a03efa70e96ee893009d6260066cde665abb8088b8cbeR21.

I admit to not really knowing if the build toolchain makes any difference - I don't think that PR changed anything else. Certainly we don't ship -pc-illumos and -sun-illumos or anything like that...

Yes, those were my thoughts as well. There is no target like -pc-illumos or -sun-illumos and considering that the build toolchain was different from the target before (and presumably it worked), I assume that changing that should not affect anything. But I am not completely sure - @jclulow might know more considering that he added that illumos target.