<div class="gmail_quote">On Wed, May 9, 2012 at 8:57 AM, Eli Friedman <span dir="ltr"><<a href="mailto:eli.friedman@gmail.com" target="_blank">eli.friedman@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="HOEnZb"><div class="h5">On Wed, May 9, 2012 at 4:30 AM, Simon Atanasyan <<a href="mailto:satanasyan@mips.com">satanasyan@mips.com</a>> wrote:<br>
> Separated from "[PATCH] Fix clang test failures on MIPS" thread.<br>
><br>
> On Thu, May 3, 2012 at 4:29 AM, Eli Friedman <<a href="mailto:eli.friedman@gmail.com">eli.friedman@gmail.com</a>> wrote:<br>
>> On Wed, May 2, 2012 at 3:04 PM, Simon Atanasyan <<a href="mailto:satanasyan@mips.com">satanasyan@mips.com</a>> wrote:<br>
>>> Clang :: Preprocessor/mmx.c<br>
>>> Clang :: Preprocessor/predefined-arch-macros.c<br>
>>><br>
>>> These test check definitions of various i386..x86-64 platforms specific<br>
>>> macros. I suggest to add "mips" to the XFAIL list for these tests.<br>
>><br>
>> For those two tests, could you just force an appropriate triple with<br>
>> -ccc-host-triple?<br>
><br>
> Yes, it is possible. The appropriate patch "tgt-force.patch" is<br>
> attached to this mail. But I'm not absolutely sure that this approach<br>
> is correct. If we pass a target triple explicitly, we check predefined<br>
> macros on that target only. If we do not pass a target triple, we<br>
> check predefined macros on each different target where we run the test<br>
> suite and we cover more cases.<br>
><br>
> But if you do not see any problem here and "tgt-force.patch" is okay,<br>
> I will commit it.<br>
<br>
</div></div>The benefit from the extra testing would be slight... better to just<br>
make it consistent. Patch is fine.</blockquote><div><br></div><div>Agreed.</div><div><br></div><div>Specifically, if we believe there might be differences from target to target, we should add explicit runs with those triples so that developers on other targets get the testing too...</div>
</div>