[webkit-dev] SVG pixel test failures (recent regressions?)

[Darin Adler]

On Jan 26, 2009, at 1:54 AM, David Hyatt wrote:

> On Jan 26, 2009, at 1:56 AM, Darin Adler wrote:
>
>> It looks to me like the pixel test failures are showing some real  
>> regressions; they must be recent ones. For example, take a look at  
>> this test.
>>
>>   http://build.webkit.org/results/trunk-mac-intel-pixel/1625/svg/text/text-align-03-b-diffs.html
>>
>> Does anyone know which code change caused that change in rendering?  
>> It sure looks like it's a new bug. Hyatt, any ideas? Anyone else?
>
> I caused the SVG pixel test regressions.  I'm still trying to figure  
> out what I broke there.

OK. Eric made it sound like this might be a bug in code in  
DumpRenderTree.

>> Here’s another curious one:
>>
>>   http://build.webkit.org/results/trunk-mac-intel-pixel/1625/fast/table/form-with-table-style-diffs.html
>>
>> Did we change the behavior of "display: table-row-group" in a way  
>> that would make things marked with that display type disappear  
>> altogether?
>
> This is a progression. Stuff that used to be incorrectly not painted  
> is now being painted.

Excellent! So we just have to generate new results for that one. I'll  
do that now.

The third category of failures seems to be related to video and  
probably has something to do with the configuration of that slave  
machine.

     -- Darin