Testing C++14 mode in libc++

Looks like we have a buildbot that says things are fine over here: http://lab.llvm.org:8011/builders/libcxx-libcxxabi-x86_64-linux-ubuntu-cxx14

I think there was actually some discussion of this on IRC earlier today. What little I saw made it sound like those were known failures if you’re using an older version if clang (such as the one that ships with xcode).

Looks like we have a buildbot that says things are fine over here:
http://lab.llvm.org:8011/builders/libcxx-libcxxabi-x86_64-linux-ubuntu-cxx14

Thanks for the link, Dan. However, I can't help but think that they ought
to be a way to systematically test this mode (without messing with
lit.site.cfg) before shipping it off... to the buildbots...
Am I to understand that the general practice here is to just rely on the
buildbots?

I think there was actually some discussion of this on IRC earlier today.
What little I saw made it sound like those were known failures if you're
using an older version if clang (such as the one that ships with xcode).

This would be impossible in my case since I am working with llvm trunk...

Looks like we have a buildbot that says things are fine over here:
http://lab.llvm.org:8011/builders/libcxx-libcxxabi-x86_64-linux-ubuntu-cxx14

Thanks for the link, Dan. However, I can't help but think that they ought
to be a way to systematically test this mode (without messing with
lit.site.cfg) before shipping it off... to the buildbots...
Am I to understand that the general practice here is to just rely on the
buildbots?

Not what I meant to imply :slight_smile: Just meant that it looks like they should be
passing.

I think there was actually some discussion of this on IRC earlier today.

What little I saw made it sound like those were known failures if you're
using an older version if clang (such as the one that ships with xcode).

This would be impossible in my case since I am working with llvm trunk...

Hmm. I just tested in my own tree, and I see them all pass just by changing
lit.site.cfg. I'm a little out of date though, r229176.

Looks like we have a buildbot that says things are fine over here:
http://lab.llvm.org:8011/builders/libcxx-libcxxabi-x86_64-linux-ubuntu-cxx14

Thanks for the link, Dan. However, I can't help but think that they ought
to be a way to systematically test this mode (without messing with
lit.site.cfg) before shipping it off... to the buildbots...
Am I to understand that the general practice here is to just rely on the
buildbots?

Not what I meant to imply :slight_smile: Just meant that it looks like they should be
passing.

Ok. Didn't think that meant to. Just wanted to clarify. :slight_smile:

I think there was actually some discussion of this on IRC earlier today.

What little I saw made it sound like those were known failures if you're
using an older version if clang (such as the one that ships with xcode).

This would be impossible in my case since I am working with llvm trunk...

Hmm. I just tested in my own tree, and I see them all pass just by
changing lit.site.cfg. I'm a little out of date though, r229176.

Ok. Strange. Let me make sure I'm not doing anything "funky" then.

What compiler are you using?

However, I can't help but think that they ought to be a way to systematically test this mode (without messing with lit.site.cfg) before shipping it off... to the buildbots...

see: http://libcxx.llvm.org/lit_usage.html

/Eric

What compiler are you using?

clang 3.7 (trunk).

>However, I can't help but think that they ought to be a way to
systematically test this mode (without messing with lit.site.cfg) before
shipping it off... to the buildbots...

see: http://libcxx.llvm.org/lit_usage.html

Thanks. Looks like

$lit --param std=c++1y ...

works but

$lit --param std=c++14 ...

does not, with "error: invalid value 'c++14' in '-std=c++14'".

Larisse Voufo <lvoufo@google.com> writes:

Roger. Thanks.