Created attachment 526565 [details] reftest log Not sure what to make of this: since it's an unexpected-pass, we must !haveTestPlugin on Android, but it passes. The log isn't particularly helpful, since for some reason I don't understand, it only contains one image, rather than the two I'd expect from a comparison.
Created attachment 526566 [details] [diff] [review] Fix v.1 But I'm not one to let my total lack of understanding get in the way of my calling it random.
Comment on attachment 526566 [details] [diff] [review] Fix v.1 This is the log, not a patch, I'm afraid.
Created attachment 526597 [details] [diff] [review] actual fix v.1 Attachments are hard. Let's go shopping!
An == test passes if the test and reference images are identical. So, an *unexpected pass* on an == test means the two images are *supposed* to be identical, they normally *aren't* due to a known bug (or in this case a known unimplemented feature), but on this particular run they *were* identical. (Enough negation for ya? :-) Anyway, under these conditions it doesn't bother dumping out both images, because they would be two copies of the same image.