[5.0.0 Release] Release Candidate 4 tagged

Hello testers,

5.0.0-rc4 was just tagged.

There were very few changes after rc3, and if nothing unexpected comes
up, this is what the final release will look like.

Please test and let me know if there are any issues.

Cheers,
Hans

Download link?

Hi,

ARM and AArch64 uploaded:
64414576baf177aa3de789343399bf6555562292
clang+llvm-5.0.0-rc4-aarch64-linux-gnu.tar.xz
d8a55e393f6277759165ad89d5b0f1b5a540c81b
clang+llvm-5.0.0-rc4-armv7a-linux-gnueabihf.tar.xz

No new failures.

Cheers,
Diana

So far it's just a tag in the svn repository. If you want the source,
you can get it from there.

In a few days, the tarballs will be uploaded to
LLVM 5.0.0 Release Candidates and I'll send a "source and
binaries available" email.

W dniu wto, 29.08.2017 o godzinie 16∶52 -0700, użytkownik Hans Wennborg
via llvm-dev napisał:

Hello testers,

5.0.0-rc4 was just tagged.

There were very few changes after rc3, and if nothing unexpected comes
up, this is what the final release will look like.

Please test and let me know if there are any issues.

Mostly green here on Gentoo. Three test problems left:

1. clang Python bindings are broken due to incomplete commit. I've
requested its revert in #34386 [1] since this is not something we can
fix without cooperation of the author, and he's not replying.

2. lldb fails a few tests:

Failing Tests (4):
    lldb :: Expr/TestCallStdStringFunction.test
    lldb :: Expr/TestCallStopAndContinue.test
    lldb :: Expr/TestCallUserAnonTypedef.test
    lldb :: Expr/TestCallUserDefinedFunction.test

3. openmp fails a few tests:

Failing Tests (2):
    libomp :: tasking/omp_taskloop_grainsize.c
    libomp :: worksharing/for/kmp_doacross_check.c

I don't think think it's really worth the effort to try to fix 2/3 now.

[1]:https://bugs.llvm.org/show_bug.cgi?id=34386

W dniu wto, 29.08.2017 o godzinie 16∶52 -0700, użytkownik Hans Wennborg
via llvm-dev napisał:

Hello testers,

5.0.0-rc4 was just tagged.

There were very few changes after rc3, and if nothing unexpected comes
up, this is what the final release will look like.

Please test and let me know if there are any issues.

Mostly green here on Gentoo. Three test problems left:

1. clang Python bindings are broken due to incomplete commit. I've
requested its revert in #34386 [1] since this is not something we can
fix without cooperation of the author, and he's not replying.

:frowning:
That commit was part of rc1-rc3 too, so I'm sad that we're only
learning about it now.

From your comment it sounds like one of the tests depends on a module

that was never committed. How come the test hasn't been failing on our
testing and on the bots?

Since this is showing so late, and it only breaks a test, not the
bindings per se, I don't think it's severe enough to block the release
on.

Thanks,
Hans

Uploaded rc4 and rc3 for SLES11. No new issues.

5560fd8bf543bf3cdcaf55caef159664bd2749f3 clang+llvm-5.0.0-rc3-linux-x86_64-sles11.3.tar.xz

effbb3a69efec75a67566d91dd4a9e685c2df15d clang+llvm-5.0.0-rc4-linux-x86_64-sles11.3.tar.xz

W dniu śro, 30.08.2017 o godzinie 15∶22 -0700, użytkownik Hans Wennborg
napisał:

> W dniu wto, 29.08.2017 o godzinie 16∶52 -0700, użytkownik Hans Wennborg
> via llvm-dev napisał:
> > Hello testers,
> >
> > 5.0.0-rc4 was just tagged.
> >
> > There were very few changes after rc3, and if nothing unexpected comes
> > up, this is what the final release will look like.
> >
> > Please test and let me know if there are any issues.
> >
>
> Mostly green here on Gentoo. Three test problems left:
>
> 1. clang Python bindings are broken due to incomplete commit. I've
> requested its revert in #34386 [1] since this is not something we can
> fix without cooperation of the author, and he's not replying.

:frowning:
That commit was part of rc1-rc3 too, so I'm sad that we're only
learning about it now.

I'm sorry. I have reported the issue a month ago and was hoping someone
would actually reply and fix it. Then I forgot about it.

From your comment it sounds like one of the tests depends on a module
that was never committed. How come the test hasn't been failing on our
testing and on the bots?

There are no bots running the Python bindings test suite, and I suppose
nobody is actually running them but people who commit new stuff to
the checks and Gentoo.

Since this is showing so late, and it only breaks a test, not the
bindings per se, I don't think it's severe enough to block the release
on.

Well, we can always fix it on our end, and hope a fix comes before
5.0.1, or we can revert it then.

Sounds good. I've made a note about it for 5.0.1.

Thanks,
Hans

Hi Hans,

There are no regressions on x86_64 Darwin. I've uploaded the package:

SHA 1: ea7da35db25b20cf65481ade67ff45bb29841b76 rc4/clang+llvm-5.0.0-rc4-x86_64-apple-darwin.tar.xz
SHA 256: 525bb26459ce0f1845c983efcb88d829709dc88d48f7c9eccd2963fdecbb9b0b rc4/clang+llvm-5.0.0-rc4-x86_64-apple-darwin.tar.xz

thanks!
vedant

Built and tested on FreeBSD 10, no changes with respect to the previous rc. I've uploaded:

SHA256 (clang+llvm-5.0.0-rc4-amd64-unknown-freebsd10.tar.xz) = d00277dbbaec71e89fcd12b92ff3d1a69af42b1d4c47a9315a46a9e4b2c656c9
SHA256 (clang+llvm-5.0.0-rc4-i386-unknown-freebsd10.tar.xz) = 5515f0e03ba4098873b39b3fac99060a622254d86d3a3b04d3b5f90df8aad9c7

-Dimitry

Windows is ready:

$ sha1sum *5.0.0-rc4*
bd13e8e9ec5c037a6b51a3e92a074b79637084e5 LLVM-5.0.0-rc4-win32.exe
c26ecf03a60825cef263f937b6f2d66c1deef006 LLVM-5.0.0-rc4-win64.exe

I've updated OpenMandriva's packages and also run a test build of AOSP
8.0.0_r4 -- looks good on both sides so far.

ttyl
bero