[Piglit] [PATCH] summary: fix support for old results file with duplicated subtests

Michel Dänzer michel at daenzer.net
Wed May 28 21:25:01 PDT 2014


On 29.05.2014 13:18, Dylan Baker wrote:
> On Thursday, May 29, 2014 13:15:54 Michel Dänzer wrote:
>> On 29.05.2014 13:09, Matt Turner wrote:
>>> On Wed, May 28, 2014 at 7:17 PM, Ilia Mirkin <imirkin at alum.mit.edu> 
> wrote:
>>>> Dylan, I'm sure you hate this, but it does seem to work for me. Not 
> sure
>>>> where you are with your fix, but this is a tool that lots of people use,
>>>> so it does need to be addressed.
>>>
>>> Does it?
>>>
>>> What's the purpose of comparing results from different versions of
>>> piglit? You're then changing multiple variables (presumably you're
>>> changing your driver too, because otherwise what are you testing?) 
> and
>>> expecting to be able to draw some kind of sensible conclusions from
>>> that?
>>
>> Yes, of course. How else would you do regular regression testing with
>> current Git of everything? It's been working fine for me.
> 
> Presumably with a single version of piglit run across multiple versions of 
> mesa.

That's not always feasible. New tests get added to piglit all the time,
and in some cases Mesa changes break piglit tests without corresponding
piglit changes. So the piglit checkout needs to be always current as
well, but you still need to be able to compare to previous results.


> Otherwise how do you know whether a change is mesa or piglit?

That's easy: I revert one of them to the previous Git snapshot, and if
that doesn't help, repeat for the other one.


-- 
Earthling Michel Dänzer            |                  http://www.amd.com
Libre software enthusiast          |                Mesa and X developer

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 234 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/piglit/attachments/20140529/84408add/attachment.sig>


More information about the Piglit mailing list