Support ResizeSynchronously for device emulation in tests.

Support ResizeSynchronously for device emulation in tests.

When device emulation is on, synchronous resize should go through emulator.

BUG=327641

[dgozman, 2014-06-10 10:30:29]

dglazkov@, pfeldman@, could you please take a look?

avi@, I usually asked jochen@ to look at emulation stuff, but while he's out,
could you please take a look as content owner?

[Avi (use Gerrit), 2014-06-10 15:34:03]

You have my LGTM from a context standpoint, conditional on getting approval from
the area experts.

[dglazkov, 2014-06-10 15:36:52]

I guess I am curious: why do we need to support this synchronous resize mode?
It's something we only do in tests, and ideally should abolish altogether,
right?

[dgozman, 2014-06-10 15:48:30]

On 2014/06/10 15:36:52, dglazkov wrote:
> I guess I am curious: why do we need to support this synchronous resize mode?
> It's something we only do in tests, and ideally should abolish altogether,
> right?

This is for https://codereview.chromium.org/308003007/, which introduces a test
where device emulation is turned on and window is resized.
It would be nice to get rid of this, but as I understand, we cannot do async
window resize in layout tests yet.

[dglazkov, 2014-06-10 15:51:36]

On 2014/06/10 at 15:48:30, dgozman wrote:
> On 2014/06/10 15:36:52, dglazkov wrote:
> > I guess I am curious: why do we need to support this synchronous resize
mode?
> > It's something we only do in tests, and ideally should abolish altogether,
> > right?
> 
> This is for https://codereview.chromium.org/308003007/, which introduces a
test where device emulation is turned on and window is resized.
> It would be nice to get rid of this, but as I understand, we cannot do async
window resize in layout tests yet.

We totally do have async window resize in layout tests! In fact, the only time
it's not async is when we use the "unfortunate mode" setting. Now, there could
be bugs that prevent you from accomplishing what's necessary for this scenario
-- we should totally fix those. Help me understand the problem a bit better?

[dgozman, 2014-06-10 15:59:07]

> We totally do have async window resize in layout tests! In fact, the only time
> it's not async is when we use the "unfortunate mode" setting. Now, there could
> be bugs that prevent you from accomplishing what's necessary for this scenario
> -- we should totally fix those. Help me understand the problem a bit better?

I've traced async resize when writing layout test, and it did nothing because
the window is not a popup nor a panel. See
https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/we....
Therefore, WebViewImpl was not resized, and test failed.

[dglazkov, 2014-06-10 16:08:18]

On 2014/06/10 at 15:59:07, dgozman wrote:
> > We totally do have async window resize in layout tests! In fact, the only
time
> > it's not async is when we use the "unfortunate mode" setting. Now, there
could
> > be bugs that prevent you from accomplishing what's necessary for this
scenario
> > -- we should totally fix those. Help me understand the problem a bit better?
> 
> I've traced async resize when writing layout test, and it did nothing because
the window is not a popup nor a panel. See
https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/we....
> Therefore, WebViewImpl was not resized, and test failed.

Got it. Hacky idea first -- can we not turn this into a popup test? Use
window.open to create the window and take it from there?

Another question is why the resize doesn't work for anything other than popup or
panel. Avi, do you happen to know?

[Avi (use Gerrit), 2014-06-10 16:10:38]

On 2014/06/10 16:08:18, dglazkov wrote:
> On 2014/06/10 at 15:59:07, dgozman wrote:
> > > We totally do have async window resize in layout tests! In fact, the only
> time
> > > it's not async is when we use the "unfortunate mode" setting. Now, there
> could
> > > be bugs that prevent you from accomplishing what's necessary for this
> scenario
> > > -- we should totally fix those. Help me understand the problem a bit
better?
> > 
> > I've traced async resize when writing layout test, and it did nothing
because
> the window is not a popup nor a panel. See
>
https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/we....
> > Therefore, WebViewImpl was not resized, and test failed.
> 
> Got it. Hacky idea first -- can we not turn this into a popup test? Use
> window.open to create the window and take it from there?
> 
> Another question is why the resize doesn't work for anything other than popup
or
> panel. Avi, do you happen to know?

Sorry, no idea.

[dgozman, 2014-06-10 16:15:42]

> Got it. Hacky idea first -- can we not turn this into a popup test? Use
> window.open to create the window and take it from there?

This may be doable, but it will probably be even more plumbing since we should
attach inspector to this popup instead of original page, and that's done in
webkit_test_runner.cc.

[dglazkov, 2014-06-10 16:20:12]

On 2014/06/10 at 16:10:38, avi wrote:
> On 2014/06/10 16:08:18, dglazkov wrote:
> > On 2014/06/10 at 15:59:07, dgozman wrote:
> > > > We totally do have async window resize in layout tests! In fact, the
only
> > time
> > > > it's not async is when we use the "unfortunate mode" setting. Now, there
> > could
> > > > be bugs that prevent you from accomplishing what's necessary for this
> > scenario
> > > > -- we should totally fix those. Help me understand the problem a bit
better?
> > > 
> > > I've traced async resize when writing layout test, and it did nothing
because
> > the window is not a popup nor a panel. See
> >
https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/we....
> > > Therefore, WebViewImpl was not resized, and test failed.
> > 
> > Got it. Hacky idea first -- can we not turn this into a popup test? Use
> > window.open to create the window and take it from there?
> > 
> > Another question is why the resize doesn't work for anything other than
popup or
> > panel. Avi, do you happen to know?
> 
> Sorry, no idea.

Here's the answer: https://code.google.com/p/chromium/issues/detail?id=2091.
Thus, the "right" solutions are:

1) Turn the test into creating a popup window and resizing it from there.
2) Expose testRunner.resize that emulates user resizing the window.