Allow tests to run with a tile cache, and optionally dump the tile cache via layerTreeAsText()

[Simon Fraser (smfr)]

Allow tests to run with a tile cache, and optionally dump the tile cache via layerTreeAsText()

[Simon Fraser (smfr)]

Created attachment 170239 [details]
Patch

[Simon Fraser (smfr)]

Created attachment 170240 [details]
Patch

[Eric Seidel (no email)]

Seems reasonable to me.

[Early Warning System Bot]

Comment on attachment 170240 [details]
Patch

Attachment 170240 [details] did not pass qt-wk2-ews (qt):
Output: http://queues.webkit.org/results/14525111

[Tim Horton]

Comment on attachment 170240 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=170240&action=review

> Source/WebCore/ChangeLog:35
> +        * page/Frame.h:
> +        * platform/graphics/GraphicsLayer.cpp:
> +        (showGraphicsLayerTree):
> +        * platform/graphics/GraphicsLayer.h:
> +        (WebCore::GraphicsLayer::tiledBacking):
> +        * platform/graphics/TiledBacking.h:
> +        * platform/graphics/ca/GraphicsLayerCA.cpp:
> +        (WebCore::GraphicsLayerCA::tiledBacking):
> +        (WebCore::GraphicsLayerCA::dumpAdditionalProperties):
> +        * platform/graphics/ca/GraphicsLayerCA.h:
> +        (GraphicsLayerCA):
> +        * platform/graphics/ca/mac/TileCache.h:
> +        (WebCore::TileCache::tileCoverageRect):
> +        (TileCache):
> +        * platform/graphics/ca/mac/TileCache.mm:
> +        (WebCore::TileCache::computeTileCoverageRect):
> +        (WebCore::TileCache::revalidateTiles):
> +        * rendering/RenderLayerCompositor.cpp:
> +        (WebCore::RenderLayerCompositor::layerTreeAsText):
> +        * testing/Internals.cpp:
> +        (WebCore::Internals::layerTreeAsText):
> +        * testing/Internals.h:
> +        * testing/Internals.idl:

Dan would be sad that there aren’t comments here.

> Source/WebCore/platform/graphics/ca/GraphicsLayerCA.cpp:2526
> +    if (m_isPageTileCacheLayer && (behavior & LayerTreeAsTextIncludePageTileCache)) {

I assume there’s no reason to actually dump all of the tile layers?

> Tools/WebKitTestRunner/TestInvocation.cpp:148
> +    return strstr(pathOrURL, "tiled-drawing/") || strstr(pathOrURL, "tiled-drawing\\");

Is this what we want? So then we can have tiled-drawing subdirectories throughout the LayoutTests tree?

> Tools/WebKitTestRunner/TestInvocation.cpp:154
> +    WKRetainPtr<WKMutableDictionaryRef> viewOptions = adoptWK(WKMutableDictionaryCreate());

Does WKRetainPtr not have the nice constructor that takes AdoptWK as the first argument, and the object as the second?

> Tools/WebKitTestRunner/mac/PlatformWebViewMac.mm:111
> +    WKRetainPtr<WKStringRef> useTiledDrawingKey(AdoptWK, WKStringCreateWithUTF8CString("TiledDrawing"));

Apparently it does :P

[Simon Fraser (smfr)]

http://trac.webkit.org/changeset/132276

[WebKit Review Bot]

Re-opened since this is blocked by bug 100189

[Csaba Osztrogonác]

Rolled out, because it broke the Qt build - http://trac.webkit.org/changeset/132298

Please don't break the build in the future and keep the contributing rules:
https://www.webkit.org/coding/contributing.html
"Keeping the tree green
...
You can watch the tree at build.webkit.org to make sure your patch builds and passes tests on all platforms.
...
Your change must at least compile on all platforms."

EWS bots and buildbots are your friends.

[Simon Fraser (smfr)]

Was rolled out in error. Ossy is rolling back in.

[Csaba Osztrogonác]

Relanded in https://trac.webkit.org/changeset/132301.

I'm so sorry for blaming you, r133275 and not r133276 was the culprit,
but the red EWS and the red tree on same revision confused me. I'll
be more careful in the future and will check the build logs better.