"recent libcxx regression" reason discovered. clang version info/folder changed. possibly accidental?

G M gmisocpp at gmail.com
Sat Mar 8 04:53:48 PST 2014


I have no opinion/preference here but didn't symlinks get removed
from an llvm/clang solution recently because on windows creating symlinks
requires the user to have certain permissions and that wasn't desirable?
I don't know if this is true or matters I just thought I would point out
that I remember seeing that recently in case it was relevant here. Sorry if
it's not.


On Sat, Mar 8, 2014 at 7:28 AM, Stellard, Thomas <Tom.Stellard at amd.com>wrote:

> Hi,
>
> I just submitted a patch which adds a symlink on both the automake and
> cmake build systems, can you test and review?
>
>
> http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20140303/100882.html
>
> Thanks,
> Tom
> ________________________________________
> From: Tom Stellard [thomas.stellard at amd.com]
> Sent: Wednesday, March 05, 2014 12:17 PM
> To: Reid Kleckner
> Cc: G M; cfe-commits at cs.uiuc.edu cfe
> Subject: Re: "recent libcxx regression" reason discovered. clang version
> info/folder changed. possibly accidental?
>
> On 03/05/2014 03:03 AM, Reid Kleckner wrote:
> > This was probably caused by r202720.  I guess that's desirable if we
> > want to update our builtin headers in a patch release, but it'd be nice
> > if we did what gcc does and symlink 3.5 to 3.5.0 or something.
> >
>
> Thanks for raising this issue.  I think we should be able to add
> a symlink from major.minor to major.minor.patch.  I will work on a patch
> for this.
>
> -Tom
>
> >
> > On Tue, Mar 4, 2014 at 3:59 AM, G M <gmisocpp at gmail.com
> > <mailto:gmisocpp at gmail.com>> wrote:
> >
> >     Hi Everyone
> >
> >     It would appear that the recent build problems that suddenly
> >     appeared for me that were manifesting through libcxx no longer
> >     building was caused by the clang headers changing directory or
> >     something like that.
> >
> >     Files like x86intrin.h were in
> >
> >     \llvm\lib\clang\3.5\include
> >
> >     but somehow for some reason changed to
> >
> >     \llvm\lib\clang\3.5.0\include
> >
> >     This may be intentional but I thought I'd call my discovery out
> >     since it might be accidental and may cause other people problems
> >     like it did me.
> >
> >     In my case It resulted in a simple hard coded script I had failing
> >     to pickup the latest versions of headers because they "moved"
> >     unexpectedly.
> >
> >     I changed my local custom script to note that ,0 change and
> >     everything builds again now.
> >
> >     Hope this info is useful to others.
> >
> >     _______________________________________________
> >     cfe-commits mailing list
> >     cfe-commits at cs.uiuc.edu <mailto:cfe-commits at cs.uiuc.edu>
> >     http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
> >
> >
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140309/0ce1431b/attachment.html>


More information about the cfe-commits mailing list