Fix invalidation of absolute sized SVG content.

Fix invalidation of absolute sized SVG content.

Minor cleanup of svg/custom/absolute-sized-content-with-resources.html
LayoutTest with no functional changes.

BUG=490725, 640273

[wkorman, 2016-08-24 17:53:16]

Description was changed from

==========
Fix invalidation of absolute sized SVG content.

BUG=490725,640273
==========

to

==========
Fix invalidation of absolute sized SVG content.

Minor cleanup of svg/custom/absolute-sized-content-with-resources.html
LayoutTest with no functional changes.

BUG=490725,640273
==========

[wkorman, 2016-08-24 17:54:56]

wkorman@chromium.org changed reviewers:
+ pdr@chromium.org, wangxianzhu@chromium.org

[wkorman, 2016-08-24 17:54:56]

I realize this is a hammer but given limited familiarity
with SVG am unsure where else to look for something that
could be more focused, open to thoughts. This does fix
this particular test.

See also https://codereview.chromium.org/2276783002/ as
another variant in review. These are both to fix SVG issues
that cropped up with the visual rect rtree patch.

[Xianzhu, 2016-08-24 18:06:23]

I would prefer https://codereview.chromium.org/2276783002/ which looks clearer
about the real problem than this CL.

https://codereview.chromium.org/2276073002/diff/1/third_party/WebKit/LayoutTe...
File third_party/WebKit/LayoutTests/TestExpectations (right):

https://codereview.chromium.org/2276073002/diff/1/third_party/WebKit/LayoutTe...
third_party/WebKit/LayoutTests/TestExpectations:63: crbug.com/598051
svg/zoom/page/zoom-foreignObject.svg [ Skip ]
I suggest to change the above Skips to Failures so that we can see their status
on flakiness dashboard.

[wkorman, 2016-08-24 18:10:35]

On 2016/08/24 18:06:23, Xianzhu wrote:
> I would prefer https://codereview.chromium.org/2276783002/ which looks clearer
> about the real problem than this CL.

Actually that other change does not fix this test. We could merge them, but in
one case, specific to foreign-objects, we need to ignore whether the previous
paint invalidation rect and new are different, so the conditionals could be hard
to read. My plan was to review after landing one, and either merge the
conditionals or keep separate if easier to read/maintain that way.

>
https://codereview.chromium.org/2276073002/diff/1/third_party/WebKit/LayoutTe...
> File third_party/WebKit/LayoutTests/TestExpectations (right):
> 
>
https://codereview.chromium.org/2276073002/diff/1/third_party/WebKit/LayoutTe...
> third_party/WebKit/LayoutTests/TestExpectations:63: crbug.com/598051
> svg/zoom/page/zoom-foreignObject.svg [ Skip ]
> I suggest to change the above Skips to Failures so that we can see their
status
> on flakiness dashboard.

Will do.

[pdr., 2016-08-24 18:17:33]

On 2016/08/24 at 18:10:35, wkorman wrote:
> On 2016/08/24 18:06:23, Xianzhu wrote:
> > I would prefer https://codereview.chromium.org/2276783002/ which looks
clearer
> > about the real problem than this CL.
> 
> Actually that other change does not fix this test. We could merge them, but in
one case, specific to foreign-objects, we need to ignore whether the previous
paint invalidation rect and new are different, so the conditionals could be hard
to read. My plan was to review after landing one, and either merge the
conditionals or keep separate if easier to read/maintain that way.

This change is really high up in the invalidation stack. Can you investigate how
all of the other svg tests are passing without this change?

[wkorman, 2016-08-24 18:21:19]

On 2016/08/24 18:17:33, pdr. wrote:
> On 2016/08/24 at 18:10:35, wkorman wrote:
> > On 2016/08/24 18:06:23, Xianzhu wrote:
> > > I would prefer https://codereview.chromium.org/2276783002/ which looks
> clearer
> > > about the real problem than this CL.
> > 
> > Actually that other change does not fix this test. We could merge them, but
in
> one case, specific to foreign-objects, we need to ignore whether the previous
> paint invalidation rect and new are different, so the conditionals could be
hard
> to read. My plan was to review after landing one, and either merge the
> conditionals or keep separate if easier to read/maintain that way.
> 
> This change is really high up in the invalidation stack. Can you investigate
how
> all of the other svg tests are passing without this change?

This could be the only one that is absolute-sized and changes the width of the
SVG element after initial layout:

wkorman@lemur:/ssd2/git/chrome/src$ find ./third_party/WebKit/LayoutTests/svg
-name "*.*html" | grep -v expected | xargs grep setProperty | grep width
./third_party/WebKit/LayoutTests/svg/as-image/svg-image-change-content-size.xhtml:
   document.getElementById("contentBox").style.setProperty("width", "600px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-content-with-resources.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-inner-svg.xhtml:     
      document.getElementById("contentBox").style.setProperty("width", "400px");
./third_party/WebKit/LayoutTests/svg/custom/object-sizing-no-width-height-change-content-box-size.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-use-on-symbol.xhtml: 
          document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-use-without-attributes-on-symbol.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/absolute-sized-content-with-resources.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-content.xhtml:       
    document.getElementById("contentBox").style.setProperty("width", "400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-image.xhtml:         
  document.getElementById("contentBox").style.setProperty("width", "400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-shadow-tree-content.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-deep-shadow-tree-content.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");
./third_party/WebKit/LayoutTests/svg/custom/relative-sized-shadow-tree-content-with-symbol.xhtml:
           document.getElementById("contentBox").style.setProperty("width",
"400px");