[llvm-commits] [llvm] r87047 - in /llvm/trunk: include/llvm/Target/TargetInstrInfo.h lib/Target/X86/X86InstrInfo.cpp lib/Target/X86/X86InstrInfo.h

[Bill Wendling]

On Nov 12, 2009, at 4:40 PM, David Greene wrote:

> On Thursday 12 November 2009 18:37, Bill Wendling wrote:
>
>>> Because it's only being used for comments at the moment and those
>>> don't
>>> need to be 100% correct (though it would be nice).
>>
>> This makes no sense to me. What do you mean "comments"? What  
>> comments?
>
> Asm comments.  The last bit still needs to be checked in (tomorrow).
>
>> I'm not happy with the state of affairs. The build bots have been
>> broken all day. Unless you can say that this patch fixes the *real*
>
> That can't be just from me.  I only started checking in at 3pm.
>
I sent a list around of the failures from when it started failing.  
They were all from you. I'm not trying to embarrass you, but that's  
what I saw.

>> problem with the bootstrap, and not some "comment" that isn't
>> "reliable" for correctness, then I'll be reverting all of your work
>> today.
>
> Yes, it fixes the "real" problem AFAICT.  Your testcase compiles.  I  
> can't
> see any other Darwin failures since I don't have access to a Darwin
> box.
>
Thank you. Please be more informative in your commit messages when  
you're fixing build failures. The message sounded like you were  
slapping a heuristic down to fix a bootstrap failure that wasn't  
reliable for correctness. Such a situation is obviously not acceptable.

*Rage subsiding*

-bw