Ho hum. I've tweaked these twice now, and had green runs on try, but on inbound they still fail. The output is a little odd too - annotations: http://mxr.mozilla.org/mozilla-central/source/layout/reftests/transform-3d/reftest.list#36 fuzzy-if(winWidget&&!layersGPUAccelerated,102,580) fuzzy-if(d2d,52,681) fails-if(Android) fuzzy-if(OSX==10.8,145,752) result: "max difference: 143, number of differing pixels: 681" I don't get it. Where is the 143 max difference coming from? 11:06:49 INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/transform-3d/scale3d-all.html | image comparison (==), max difference: 143, number of differing pixels: 681 11:06:49 INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/transform-3d/scale3d-all-separate.html | image comparison (==), max difference: 143, number of differing pixels: 681

rgb color difference

(In reply to Ryan VanderMeulen [:RyanVM] from comment #1) > rgb color difference That's the root cause of the failure, but the annotations instruct the harness to ignore differences <= 681 pixels. In the output it says "max difference: 143", I don't see any rule that specifies a maximum difference of 143 pixels.

I think I see the problem. the first number in the fuzzy-if rule is not a min, it's a 'fuzzy_max_delta'.

Pushed this to try, lets see how they come out.

Comment on attachment 727170 [details] [diff] [review] patch This finally got it. I was confused about the use of the first value in fuzzy-if. https://tbpl.mozilla.org/?tree=Try&rev=1e5d95ee9076&showall=1

Comment on attachment 727170 [details] [diff] [review] patch Review of attachment 727170 [details] [diff] [review]: ----------------------------------------------------------------- Looks good to me although I don't think I have the authority to r+ it. But if you think so that's good enough for me ;)