Do not navigate "other" windows to about:blank in OnReset.

Do not navigate "other" windows to about:blank in OnReset.

In crrev.com/14046012 we started to iterate over all RenderViews and
navigating them to about:blank to trigger their unload event handlers
while the main window is still present.  This code is broken when OOPIFs
are enabled (i.e. we cannot navigate a *remote* RenderView / RenderFrame).

This CL basically reverts crrev.com/14046012 and instead puts the burden
on test authors, to make sure that their unload handlers can robustly
handle being executed when the main window is gone.  This makes the
test harness work better in presence of OOPIFs.

BUG=583089
Committed: https://crrev.com/0302144e0addc087dc56bb67139ec171c210e353
Cr-Commit-Position: refs/heads/master@{#374739}

[Łukasz Anforowicz, 2016-02-02 00:38:00]

lukasza@chromium.org changed reviewers:
+ jochen@chromium.org

[Łukasz Anforowicz, 2016-02-02 00:38:00]

jochen@, could you please take a look? (since this effectively reverts your
crrev.com/14046012 from 2013;  FWIW, the blink trybots are happy with patchset
1)

[Łukasz Anforowicz, 2016-02-02 20:39:08]

FWIW, I cannot repro the linux_blink_dbg failures locally with:

$ DISPLAY=:20 third_party/WebKit/Tools/Scripts/run-webkit-tests -t Debug -v
--iterations=20 --test-list=$HOME/scratch/testlist

$ cat ~/scratch/testlist                                                        
                                                                    
animations/change-one-anim.html
http/tests/navigation/lockedhistory-iframe.html
fast/dom/shadow/svg-style-in-shadow-tree-crash.html
fast/events/touch/gesture/pad-gesture-fling.html
virtual/scroll_customization/fast/events/touch/gesture/pad-gesture-fling.html
fast/encoding/char-encoding.html
fast/dom/Geolocation/cached-position-called-once.html
virtual/smoothscrolling/fast/events/touch/gesture/pad-gesture-fling.html
fast/dom/Window/window-function-frame-getter-precedence.html
fast/js/regress/string-cons-tower.html

OTOH, I was able to repro the failure below without the CL / at ToT:
- failed 20 out of 20 times: fast/dom/shadow/scoped-events-by-ua-stopped.html

[Łukasz Anforowicz, 2016-02-02 21:54:06]

Actually, I see that some failures in
https://build.chromium.org/p/tryserver.blink/builders/linux_blink_dbg/builds/...
also happened "without the patch" - this leaves us with only 5 failures that are
supposedly specific to this CL, but still won't repro on my machine.

$ DISPLAY=:20 third_party/WebKit/Tools/Scripts/run-webkit-tests -t Debug -v
--iterations=100 --test-list=$HOME/scratch/testlist
... everything passed ...
$ cat $HOME/scratch/testlist                                                    
                                                                
fast/encoding/char-encoding.html
fast/js/regress/string-cons-tower.html
fast/forms/submit-to-blank-multiple-times.html
virtual/smoothscrolling/fast/events/touch/gesture/pad-gesture-fling.html
fast/dom/shadow/svg-style-in-shadow-tree-crash.html

[jochen (gone - plz use gerrit), 2016-02-03 13:30:38]

lgtm

[jochen (gone - plz use gerrit), 2016-02-03 13:32:16]

well, actually, lgtm once it passes tests :)

To reproduce test failures, I recommend to take the output from the bots,
extract the list of tests that ran on the worker where you see the failure up
until the failure, and then bisect that list down.

[Łukasz Anforowicz, 2016-02-04 21:16:56]

lukasza@chromium.org changed reviewers:
+ mkwst@chromium.org

[Łukasz Anforowicz, 2016-02-04 21:16:57]

Mike, could you please do an OWNERS review?  (I think this can happen in
parallel / doesn't need to wait until I investigate the failures seen on
linux_blink_dbg bot - some of them are already understood as unrelated to this
CL [see below for more info]).

On 2016/02/03 13:32:16, jochen wrote:
> well, actually, lgtm once it passes tests :)
> 
Right :-)

> To reproduce test failures, I recommend to take the output from the bots,
> extract the list of tests that ran on the worker where you see the failure up
> until the failure, and then bisect that list down.

Good idea.  I've taken around 50 tests from before each failure (bot runs things
in parallel, but I've tried to always include 5+ tests on the same worker) and
run all of the locally: once with --iterations=10 and once with --iterations=3 +
--child-processes=1.  Unfortunately, the failures did not repro locally.

Also note that:

1. Some of the timed-out tests are just slow and need to be marked as such
(crrev.com/1668703006 should take care of that)

2. linux_blink_dbg is the only failing bot (and this is normally not included by
the CQ).

[jochen (gone - plz use gerrit), 2016-02-05 14:51:23]

i'd run content_shell directly (use --run-layout-tests test1 test2 test3...) for
http tests, you'll have to fire up the test servers using run-blink-httpd and
replace http/tests with http://localhost:8000/ or something (I don't recall the
default value)

[Łukasz Anforowicz, 2016-02-10 17:06:31]

I plan to do a dry-run against CQ and land this CL later today:
1. After rebasing the tests are passing even on linux_blink_dbg (most likely
because of marking some tests as slow in crrev.com/1668703006).
2. I didn't realize that jochen@ is //content OWNER and that I don't necessarily
need an extra sign-off from //content/shell OWNERS.

Please speak up if you have any concerns.

[Łukasz Anforowicz, 2016-02-10 17:06:57]

The CQ bit was checked by lukasza@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-02-10 17:09:26]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1656033004/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1656033004/40001

[commit-bot: I haz the power, 2016-02-10 18:14:34]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-02-10 18:14:35]

Dry run: This issue passed the CQ dry run.

[Łukasz Anforowicz, 2016-02-10 21:43:39]

The CQ bit was checked by lukasza@chromium.org

[Łukasz Anforowicz, 2016-02-10 21:43:39]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org
Link to the patchset: https://codereview.chromium.org/1656033004/#ps40001
(title: "Rebasing the expectations file.")

[commit-bot: I haz the power, 2016-02-10 21:46:51]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1656033004/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1656033004/40001

[commit-bot: I haz the power, 2016-02-10 21:58:11]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-02-16 22:31:28]

Description was changed from

==========
Do not navigate "other" windows to about:blank in OnReset.

In crrev.com/14046012 we started to iterate over all RenderViews and
navigating them to about:blank to trigger their unload event handlers
while the main window is still present.  This code is broken when OOPIFs
are enabled (i.e. we cannot navigate a *remote* RenderView / RenderFrame).

This CL basically reverts crrev.com/14046012 and instead puts the burden
on test authors, to make sure that their unload handlers can robustly
handle being executed when the main window is gone.  This makes the
test harness work better in presence of OOPIFs.

BUG=583089
==========

to

==========
Do not navigate "other" windows to about:blank in OnReset.

In crrev.com/14046012 we started to iterate over all RenderViews and
navigating them to about:blank to trigger their unload event handlers
while the main window is still present.  This code is broken when OOPIFs
are enabled (i.e. we cannot navigate a *remote* RenderView / RenderFrame).

This CL basically reverts crrev.com/14046012 and instead puts the burden
on test authors, to make sure that their unload handlers can robustly
handle being executed when the main window is gone.  This makes the
test harness work better in presence of OOPIFs.

BUG=583089

Committed: https://crrev.com/0302144e0addc087dc56bb67139ec171c210e353
Cr-Commit-Position: refs/heads/master@{#374739}
==========

[commit-bot: I haz the power, 2016-02-16 22:31:28]

Patchset 3 (id:??) landed as
https://crrev.com/0302144e0addc087dc56bb67139ec171c210e353
Cr-Commit-Position: refs/heads/master@{#374739}