Bug 63190 – compositing/layer-creation/spanOverlapsCanvas.html failing on Windows since it was added

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Adam Roben (:aroben)]

compositing/layer-creation/spanOverlapsCanvas.html has been failing on Windows since it was added in r89474.

http://build.webkit.org/results/Windows%207%20Release%20(Tests)/r89476%20(14034)/results.html

[Matthew Delaney]

Created attachment 98249 [details]
Patch

[Matthew Delaney]

Committed r89488: <http://trac.webkit.org/changeset/89488>

[Matthew Delaney]

Reopening to fix layout test to *actually* use the Layer Tree Text. +rebaselining the results for mac and windows as well.

[Matthew Delaney]

Created attachment 98266 [details]
Patch

[James Robinson]

Comment on attachment 98266 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=98266&action=review

> LayoutTests/compositing/layer-creation/spanOverlapsCanvas.html:15
> +                window.layoutTestController.dumpAsText();

dumpAsText(true) to get it to produce pixel output as well

> LayoutTests/platform/mac/compositing/layer-creation/spanOverlapsCanvas-expected.txt:1
> +(GraphicsLayer (bounds 800.00 600.00) (children 1 (GraphicsLayer (bounds 800.00 600.00) (children 2 (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) ) ) ) )

can this graphics layer tree go alongside the test itself?  it should be reasonably cross-platform, i think

also, if you put this into a <pre> instead of a <span> then the newlines will be preserved, which is much more readable

[Matthew Delaney]

> > LayoutTests/platform/mac/compositing/layer-creation/spanOverlapsCanvas-expected.txt:1
> > +(GraphicsLayer (bounds 800.00 600.00) (children 1 (GraphicsLayer (bounds 800.00 600.00) (children 2 (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) ) ) ) )
> 
> can this graphics layer tree go alongside the test itself?  it should be reasonably cross-platform, i think
Go alongside the test? Not sure what you mean. And I know for sure that it differs for some of our ports, so the output of it is not cross platform.

> 
> also, if you put this into a <pre> instead of a <span> then the newlines will be preserved, which is much more readable

Good call.

[James Robinson]

(In reply to comment #6)
> > > LayoutTests/platform/mac/compositing/layer-creation/spanOverlapsCanvas-expected.txt:1
> > > +(GraphicsLayer (bounds 800.00 600.00) (children 1 (GraphicsLayer (bounds 800.00 600.00) (children 2 (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) (GraphicsLayer (position 8.00 8.00) (bounds 100.00 50.00) (drawsContent 1) ) ) ) ) )
> > 
> > can this graphics layer tree go alongside the test itself?  it should be reasonably cross-platform, i think
> Go alongside the test? Not sure what you mean. And I know for sure that it differs for some of our ports, so the output of it is not cross platform.
> 

I mean put the -expected.txt next to the test.  I know that mac-snowleopard (and below) will differ and need its own expectation, but I think that this output is the most common going forward.  I'm pretty sure chromium will produce the same layer tree, for instance.

[Matthew Delaney]

Created attachment 98273 [details]
Patch

[Matthew Delaney]

Created attachment 98279 [details]
Patch

[James Robinson]

Comment on attachment 98279 [details]
Patch

R=me. the move + edit is confusing the review tool and the changelog script, but i think it'll apply OK

[Matthew Delaney]

Committed r89511: <http://trac.webkit.org/changeset/89511>