r365030 - Make a buildbot using a buggy gcc happy

[Richard Smith via cfe-commits]

I'll commit the change below once my testing finishes :)

On Mon, 8 Jul 2019 at 12:40, Kristóf Umann via cfe-commits <
cfe-commits at lists.llvm.org> wrote:

> Noted, thanks! Gabor, could you please fix this?
>
> On Mon, 8 Jul 2019, 21:37 Richard Smith, <richard at metafoo.co.uk> wrote:
>
>> This is in any case the wrong fix. The code *is* wrong, for the reason
>> this compiler is reporting.
>>
>> The correct fix is to declare the explicit specializations in the header
>> file:
>>
>> template <> void CFGDominatorTreeImpl<true>::anchor();
>> template <> void CFGDominatorTreeImpl<false>::anchor();
>>
>> Clang will tell you to do this under -Wundefined-func-template (which we
>> haven't turned on by default because people get this wrong too often...).
>>
>> On Mon, 8 Jul 2019 at 12:29, JF Bastien via cfe-commits <
>> cfe-commits at lists.llvm.org> wrote:
>>
>>> Kristof,
>>>
>>> It looks like your fix didn’t address all the bots:
>>>
>>> /Users/buildslave/jenkins/workspace/clang-stage2-coverage-R/llvm/tools/clang/lib/Analysis/Dominators.cpp:14:48:
>>> error: explicit specialization of 'anchor' after instantiation void
>>> CFGDominatorTreeImpl</*IsPostDom=*/true>::anchor() {} ^
>>> /Users/buildslave/jenkins/workspace/clang-stage2-coverage-R/llvm/tools/clang/include/clang/Analysis/Analyses/Dominators.h:225:3:
>>> note: implicit instantiation first required here
>>> ControlDependencyCalculator(CFG *cfg) ^
>>>
>>> Can you please address the issue?
>>>
>>> http://green.lab.llvm.org/green/job/clang-stage2-coverage-R/4153/consoleFull
>>>
>>> Thanks,
>>>
>>> JF
>>>
>>>
>>> On Jul 3, 2019, at 5:06 AM, Kristof Umann via cfe-commits <
>>> cfe-commits at lists.llvm.org> wrote:
>>>
>>> Author: szelethus
>>> Date: Wed Jul  3 05:06:10 2019
>>> New Revision: 365030
>>>
>>> URL: http://llvm.org/viewvc/llvm-project?rev=365030&view=rev
>>> Log:
>>> Make a buildbot using a buggy gcc happy
>>>
>>> When specializing a template in a namespace, it has to be in a namespace
>>> block, else gcc will get confused. Hopefully this fixes the issue.
>>>
>>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56480
>>>
>>> Modified:
>>>    cfe/trunk/lib/Analysis/Dominators.cpp
>>>
>>> Modified: cfe/trunk/lib/Analysis/Dominators.cpp
>>> URL:
>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Analysis/Dominators.cpp?rev=365030&r1=365029&r2=365030&view=diff
>>>
>>> ==============================================================================
>>> --- cfe/trunk/lib/Analysis/Dominators.cpp (original)
>>> +++ cfe/trunk/lib/Analysis/Dominators.cpp Wed Jul  3 05:06:10 2019
>>> @@ -8,10 +8,12 @@
>>>
>>> #include "clang/Analysis/Analyses/Dominators.h"
>>>
>>> -using namespace clang;
>>> +namespace clang {
>>>
>>> template <>
>>> -void clang::CFGDominatorTreeImpl</*IsPostDom=*/true>::anchor() {}
>>> +void CFGDominatorTreeImpl</*IsPostDom=*/true>::anchor() {}
>>>
>>> template <>
>>> -void clang::CFGDominatorTreeImpl</*IsPostDom=*/false>::anchor() {}
>>> +void CFGDominatorTreeImpl</*IsPostDom=*/false>::anchor() {}
>>> +
>>> +} // end of namespace clang
>>>
>>>
>>> _______________________________________________
>>> cfe-commits mailing list
>>> cfe-commits at lists.llvm.org
>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>>
>>>
>>> _______________________________________________
>>> cfe-commits mailing list
>>> cfe-commits at lists.llvm.org
>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>>
>> _______________________________________________
> cfe-commits mailing list
> cfe-commits at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20190708/8442705b/attachment-0001.html>