[PATCH] D25683: [docs] Increase minimum supported GCC version for building LLVM to 4.8

[Aaron Ballman via llvm-commits]

On Tue, Oct 18, 2016 at 2:00 PM, Aaron Ballman <aaron at aaronballman.com> wrote:
> On Tue, Oct 18, 2016 at 1:52 PM, Teresa Johnson <tejohnson at google.com> wrote:
>>
>>
>>
>> On Tue, Oct 18, 2016 at 10:29 AM, Aaron Ballman <aaron at aaronballman.com> wrote:
>>>
>>> It turns out that there *may* be a problem with dropping GCC 4.7 right
>>> now. It seems that our documentation builder (which is not a public
>>> bot, unfortunately) that is used for building the public-facing
>>> attribute and compiler option documents is on a machine running GCC
>>> 4.7. I've alerted the person maintaining the bot about it, but I'm not
>>> certain how soon that builder will be switched over (it may already be
>>> in the process of switching, for all I know).
>>>
>>> I apologize for not recognizing this earlier, but it turns out that a
>>> recent commit broke that bot this morning, which is how I realized
>>> that the bot is still on GCC 4.7.
>>
>>
>> Ok do we need to revert this change? If so, how long would it take to get the bot upgraded to gcc 4.8? I am surprised the bot has been working though - this change was motivated by multiple reports of clang and LLVM not building with gcc 4.7. Or is it just building documentation?
>
> I think we may need to revert this change, but I'll know for sure
> tomorrow when the bot gives me a status report.

I woke up this morning to a status report:

Updated to revision 284575.
CMake Error at cmake/modules/CheckCompilerVersion.cmake:12 (message):
  Host GCC version must be at least 4.8!
Call Stack (most recent call first):
  cmake/config-ix.cmake:14 (include)
  CMakeLists.txt:517 (include)

-- Configuring incomplete, errors occurred!
See also "/opt/llvm/build.attributes/CMakeFiles/CMakeOutput.log".
See also "/opt/llvm/build.attributes/CMakeFiles/CMakeError.log".

So it looks like this patch effectively disables generation of the
public documentation for attributes and compiler options. We don't
change those so frequently that a revert is critical, but unless we
can get the bot updated reasonably soon, this would mean our public
docs for those two things are basically in stasis.

I will continue to push for the bot upgrade, but I am not able to
control the process, only request it.

~Aaron

> I'm a consumer of this
> bot, but not the producer of it. I don't have access to the bot, but
> it sends me an automatic status update after it builds the
> documentation, which is done on a daily schedule.
>
> I've emailed Tanya (who helped me get this builder set up) about
> upgrading the bot, but I've not heard back on the status or a
> timeline. AFAIK, the bot is only trying to build the clang-tblgen tool
> (and what it relies on), which probably explains why it hasn't broken
> until yesterday.
>
> ~Aaron
>
>>
>> Thanks,
>> Teresa
>>
>>
>>>
>>>
>>> ~Aaron
>>>
>>>
>>> On Tue, Oct 18, 2016 at 12:40 PM, Renato Golin via llvm-commits
>>> <llvm-commits at lists.llvm.org> wrote:
>>> > rengolin accepted this revision.
>>> > rengolin added a comment.
>>> > This revision is now accepted and ready to land.
>>> >
>>> > In https://reviews.llvm.org/D25683#573000, @jlebar wrote:
>>> >
>>> >> If someone really wants to use gcc 4.7, nothing is stopping them from going into cmake and commenting out this line.  :)
>>> >
>>> >
>>> > Good point. :)
>>> >
>>> > LGTM. Thanks!
>>> >
>>> >
>>> > https://reviews.llvm.org/D25683
>>> >
>>> >
>>> >
>>> > _______________________________________________
>>> > llvm-commits mailing list
>>> > llvm-commits at lists.llvm.org
>>> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>>
>>
>>
>>
>> --
>> Teresa Johnson | Software Engineer | tejohnson at google.com | 408-460-2413