[llvm-commits] [llvm] r169886 - /llvm/trunk/lib/MC/MCAssembler.cpp

[David Blaikie]

On Wed, Jan 9, 2013 at 4:06 AM, Timur Iskhodzhanov <timurrrr at google.com> wrote:
> 2013/1/3 Eli Bendersky <eliben at google.com>:
>>> Is there any way we can help the Clang/LLVM test/bot infrastructure so that
>>> the Win breakages are more obvious?
>>> Set up more bots?
>>> Set up weekly/daily/hourly runs of the C-subset of SPEC CPU tests?
>>>
>>> Timur Iskhodzhanov,
>>> Google Russia
>>>
>>
>> Michael's r171437 should have solved this (by moving the instruction
>> emission on COFF to the common code in MCObjectStreamer). Timur, do
>> you still see failures after his commit?
> My build looks healthy now, thanks!
>
>> I wish we had some simple regression tests to reproduce such problems
>> in the future.
> +1
> How come this is not tested regularly?

(Coming in a little late to this party, so I may be missing some context)

LLVM(&Clang) testing on Windows is pretty much only done by Takumi. We
haven't got anything running in the official LLVM lab.

I'm not sure if Takumi would want to have his buildmaster pointing at
new slaves contributed/hosted in lab.llvm.org (or if that's even an
option) or how else we could use more resources. (also not sure which
scenarios he's got covered/where the holes are that are relevant to
this thread)

- David