[webkit-dev] How to deal with 1-pixel differences in reftests ?

[Myles C. Maxfield]

I believe that any way to specify fuzziness should not treat the following cases the same:

- 1 pixel being 100% wrong
- All pixels being 1% wrong

There are certain cases where one of those should cause a failure, but the other shouldn't.

--Myles
> On Nov 18, 2015, at 10:44 PM, Alexey Proskuryakov <ap at webkit.org> wrote:
> 
> 
>> 18 нояб. 2015 г., в 11:50, Simon Fraser <simon.fraser at apple.com <mailto:simon.fraser at apple.com>> написал(а):
>> 
>> There are some well-understood reasons why a test might not exactly match its reference. One is that the test uses compositing layers to do clipping, but the reference just clips with drawing, and these are not expected to give a pixel-perfect match.
>> 
>> I proposed a way to allow a test to specify a custom tolerance in https://bugs.webkit.org/show_bug.cgi?id=149828 <https://bugs.webkit.org/show_bug.cgi?id=149828>. If we had this, it would allow me to fix 142258 <https://bugs.webkit.org/show_bug.cgi?id=142258> and 146523 <https://bugs.webkit.org/show_bug.cgi?id=146523>.
> 
> Bug 142258 also serves as an example for why we shouldn't do this. Both known reasons for it are actual bugs that needed to be discovered, and need to be fixed.
> 
> What are the causes for flakiness in bug 146523?
> 
> - Alexey
> 
> _______________________________________________
> webkit-dev mailing list
> webkit-dev at lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-dev/attachments/20151130/849e8b23/attachment.html>