[libvirt] [PATCH] Read PCI class from sysfs class file instead of config space.

[Laine Stump]

On 12/11/2013 04:46 PM, Thadeu Lima de Souza Cascardo wrote:
> On Wed, Dec 11, 2013 at 04:30:48PM +0200, Laine Stump wrote:
>> Aside from the above lack of error logging, as Michal suggested a test
>> case would be very useful.
> I will work on that as well. I guess this should be a different commit,
> and could even be applied before this one. I'll test
> virPCIDeviceIsAssignable, which is the public function here, and provide
> two test cases, one that works without this patch, and one that doesn't.
>
> Does that work? Any suggestions?

Whenever possible, a test case for new/different functionality (or to
prove that the code changes didn't break some other desired existing
behavior) should be in the same commit as the new functionality
(certainly not in a commit *before* the new functionality, as that would
give us a version in git that failed make check). If you wanted you
could make the test that "works without this patch" a separate commit,
but I don't think that's necessary.