Stretched screen (briefly) after rotating

Status changed to 'Confirmed' because the bug affects multiple users.

The attached branch solves most of the issue. But it's wrong to say the whole bug is solved by it.

Additional suggested improvement: Try animating the resize linearly in proportion to the rotation angle/progress. That way even if the app is slightly behind at the end, it will only be slightly behind with almost the right new dimensions already.

dandrader has just about convinced me that Mir is also broken. If an app receives a resize event and only redraws once per resize event, then the delayed nature of the buffer queue might mean the app stops resizing (answers the last resize event) still using the old buffer dimensions.

We should keep resending resize events till their dimensions match that of the buffer the client last submitted.

Although fixing bug 1288021 well enough might negate the need for an additional Mir fix here.

We also have a reasonable workaround (arguably a fix) planned for clients like qtubuntu which reduces the need for a Mir fix.

lp:~dandrader/qtubuntu/resizeCatchUp solves the problem we have currently that by symply rotating an app from portrait to landscape (or vice-versa) you can get the app stretched. Because it didn't consume enough buffers to reach the first one with the new size that was promised on the ResizeEvent

The remaining issue is that when you press the power button to turn the display off and then again to turn it back on, MirSurfaceItem gets resized in the scene but the underlying mir surface does not get resized at all. Likely because the apps was already unfocused and getting suspended by the time the resize takes place

Note that the symptoms have changed slightly since this bug was originally reported.

Status changed to 'Confirmed' because the bug affects multiple users.