Bugzilla

We have a running scroll animation on the compositor, we sample it, it produces reasonable (and non-zero) changes every time we sample it but the compositor still determines nothing has changed and the invalid region for the composite is empty. (The same problem happens with and without wr, so I'm debugging the non-wr version because it is slightly easier.)

It looks like it supposed to work by AsyncCompositionManager::TransformShadowTree first calling Sample on all running animations, and then AsyncCompositionManager::ApplyAsyncContentTransformToTree sets a new base shadow transform using the update values, and then ComputeDifferences with the cloned layer properties is supposed to generate an invalid region to composite, except that region is empty sometimes when the animation, and hence the new transform, changed by reasonable numbers (not possible they rounded/snapped down to zero). Not sure why yet, or if it's some other problem causing that.

(edited later: this comment seems to be incorrect)

We have a running scroll animation on the compositor, we sample it, it produces reasonable (and non-zero) changes every time we sample it but the compositor still determines nothing has changed and the invalid region for the composite is empty. (The same problem happens with and without wr, so I'm debugging the non-wr version because it is slightly easier.)

It looks like it supposed to work by AsyncCompositionManager::TransformShadowTree first calling Sample on all running animations, and then AsyncCompositionManager::ApplyAsyncContentTransformToTree sets a new base shadow transform using the update values, and then ComputeDifferences with the cloned layer properties is supposed to generate an invalid region to composite, except that region is empty sometimes when the animation, and hence the new transform, changed by reasonable numbers (not possible they rounded/snapped down to zero). Not sure why yet, or if it's some other problem causing that.