ship rust analyzer

[matklad]

This successfully builds rust-analyzer as a part of rust repo. I haven't yet added required changes to dist.rs -- seems like I just have to copy-paste quite a bit of code I don't really understand :-)

[rust-highfive]

⚠️ Warning ⚠️

• These commits modify submodules.

[Mark-Simulacrum]

Yeah, the dist.rs modification strategy is like that for everyone I think :)

But this looks like a good start to me -- let me know if you have specific questions that I can help answer; once we think dist.rs is in good shape a try build should produce sufficient artifacts for testing on linux (x86_64) and once that's working we can land it! 🎉

[matklad]

Ready for review @Mark-Simulacrum I think.

[Mark-Simulacrum]

All of those shouldn't affect a try build I think so let's see if this is doing what it's supposed to, @bors try

[bors]

⌛ Trying commit 22c9db138d7dda91c127c241919acc50a5921dcf with merge 37d6cce42752fb1b3ae7414d5c142c99dffc53f5...

[ehuss]

Will this be tracked in toolstate in the future?

[Mark-Simulacrum]

@ehuss I don't think so -- rust-analyzer builds with a stable compiler today so unless that changes in the future (and I think we don't want it to) it shouldn't break on changes to the compiler so there's no reason for toolstate really.

[Mark-Simulacrum]

Heh, looks like rust-analyzer failed to build on our dist builders. I'm not sure whether this is "expected" -- I guess maybe we just need to install inotify? There's also the lutimes warning which is somewhat concerning.

It does look like force pushing kills bors from notifying that the try build is done (but the try build still completes).

"-Wl,-rpath,$ORIGIN/../lib"�[0m
2020-06-04T17:18:41.4210399Z �[0m  �[0m�[0m�[1m�[38;5;12m= �[0m�[0m�[1mnote�[0m�[0m: /checkout/obj/build/x86_64-unknown-linux-gnu/stage1-tools/x86_64-unknown-linux-gnu/release/deps/libfiletime-cd4b67c5479ae270.rlib(filetime-cd4b67c5479ae270.filetime.4z3loc0p-cgu.4.rcgu.o): In function `filetime::imp::utimes::set_times::hb28e25c679b44a1a':�[0m
2020-06-04T17:18:41.4211763Z �[0m          filetime.4z3loc0p-cgu.4:(.text._ZN8filetime3imp6utimes9set_times17hb28e25c679b44a1aE+0x287): warning: warning: lutimes is not implemented and will always fail�[0m
2020-06-04T17:18:41.4215999Z �[0m          /checkout/obj/build/x86_64-unknown-linux-gnu/stage1-tools/x86_64-unknown-linux-gnu/release/deps/libinotify-cd7a29090309d6ed.rlib(inotify-cd7a29090309d6ed.inotify.bxraksv4-cgu.3.rcgu.o): In function `inotify::inotify::Inotify::init::he432cafdcf6fb733':�[0m
2020-06-04T17:18:41.4217235Z �[0m          inotify.bxraksv4-cgu.3:(.text._ZN7inotify7inotify7Inotify4init17he432cafdcf6fb733E+0x10): undefined reference to `inotify_init1'�[0m
2020-06-04T17:18:41.4218077Z �[0m          clang-9: error: linker command failed with exit code 1 (use -v to see invocation)�[0m
2020-06-04T17:18:41.4218761Z �[0m          �[0m
2020-06-04T17:18:41.4218902Z 
2020-06-04T17:18:41.4243114Z �[0m�[1m�[38;5;9merror�[0m�[0m�[1m: aborting due to previous error�[0m
2020-06-04T17:18:41.4243571Z 
2020-06-04T17:18:41.4450968Z �[0m�[0m�[1m�[31merror�[0m�[1m:�[0m could not compile `rust-analyzer`.

[mati865]

@Mark-Simulacrum you cannot just install inotify.
inotify_init1 comes from GLIBC (2.9 should be fine?) but Linux dist uses GLIBC 2.5 for compatibility reasons.

[matklad]

Yup, as mati865 said, it seems like our file watching library requires glibc 2.9. I don't think it's reasonable to bump glibc version we use to build the rust toolchain, so we'd have to change something on rust-analyzer side. This is somewhat sad, because our file-watching functionality is known to be broken in various ways anyway (by default, we use editor-side file watching if that's available).

[lnicola]

CC #62516. I think we'd be able to drop support for CentOS 5 in December at earliest.

[matklad]

blocked on resolving the issue in Rust inotify wrappar: rust-lang/rust-analyzer#4224 (comment).

[bors]

⌛ Testing commit 92b5dd76bc30e985cfc989438838e7d9cda16b93 with merge fcdd79ab164987b04c0e6a4a1e0adcaf6040cb2c...

[bors]

💔 Test failed - checks-actions

[matklad]

So, this failed on power because we use AtomicU64 in salsa. Let me check if this is fixable in salsa...

[matklad]

Updated salsa and checked that rust-analyzer cargo checks locally with powerpc target after update

@bors r=Mark-Simulacrum

[bors]

📌 Commit 058c1b6 has been approved by Mark-Simulacrum

[bors]

⌛ Testing commit 058c1b6 with merge 743a9fab5f5e724bf0b5364e802f3d1f8673e079...

[bors]

💔 Test failed - checks-actions

[matklad]

Some cargo test failed. Seems unrelated?

[Manishearth]

@bors retry

seems like a network issue

[bors]

⌛ Testing commit 058c1b6 with merge 0cd7ff7...

[bors]

☀️ Test successful - checks-actions, checks-azure
Approved by: Mark-Simulacrum
Pushing 0cd7ff7 to master...