Marking a few extra layout tests as slow.

Marking a few extra layout tests as slow.

When running tryjobs against crrev.com/1656033004 and
crrev.com/1660653005 I noticed flaky timeouts on linux_blink_dbg that
didn't seem related to my changes.  Running the tests locally revealed
that they are indeed a bit slow and that we should mark them as such (to
reduce tryjob flakiness).

Below results are from my local machine (Z420, Linux, Debug build) after
running tests 10 times (via --iterations=10), gathering their times (via
--timing and --verbose) and discarding slowest and fastest time (to get
rid of outliers):
- fast/css/should-not-insert-stylesheet-into-detached-document.html: 2.5s - 2.6s
- fast/dom/shadow/svg-style-in-shadow-tree-crash.html: 1.9s - 2.5s
- fast/encoding/char-encoding.html: 1.6s - 2.2s
- fast/frames/sandboxed-iframe-navigation-targetlink.html: 1.7s - 2.2s
- fast/html/marquee-destroyed-without-removed-from-crash.html: 1.8s - 2.7s
- fast/js/regress/string-cons-tower.html: 1.7s - 2.5s
- imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html: 2s - 2.7s
- virtual/pointerevent/fast/events/tabindex-focus-blur-all.html: 2.7s - 3s

BUG=24182, 577381, 477150
Committed: https://crrev.com/8af8c1542f97215f2acad97cc5dbccfa81889c89
Cr-Commit-Position: refs/heads/master@{#373647}

[Łukasz Anforowicz, 2016-02-04 19:37:55]

Description was changed from

==========
Marking a few extra layout tests as slow.

When running tryjobs against crrev.com/1656033004 and
crrev.com/1660653005 I noticed flaky timeouts on linux_blink_dbg that
didn't seem related to my changes.  Running the tests locally revealed
that they are indeed a bit slow and that we should mark them as such (to
reduce tryjob flakiness).

Below results are from my local machine (Z420, Linux, Debug build) after
running tests 10 times and discarding slowest and fastest time (to get
rid of outliers):
- fast/css/should-not-insert-stylesheet-into-detached-document.html: 2.5s - 2.6s
- fast/dom/shadow/svg-style-in-shadow-tree-crash.html: 1.9s - 2.5s
- fast/encoding/char-encoding.html: 1.6s - 2.2s
- fast/frames/sandboxed-iframe-navigation-targetlink.html: 1.7s - 2.2s
- fast/html/marquee-destroyed-without-removed-from-crash.html: 1.8s - 2.7s
- fast/js/regress/string-cons-tower.html: 1.7s - 2.5s
- imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html: 2s -
2.7s
- virtual/pointerevent/fast/events/tabindex-focus-blur-all.html: 2.7s - 3s

BUG=24182, 577381
==========

to

==========
Marking a few extra layout tests as slow.

When running tryjobs against crrev.com/1656033004 and
crrev.com/1660653005 I noticed flaky timeouts on linux_blink_dbg that
didn't seem related to my changes.  Running the tests locally revealed
that they are indeed a bit slow and that we should mark them as such (to
reduce tryjob flakiness).

Below results are from my local machine (Z420, Linux, Debug build) after
running tests 10 times (via --iterations=10), gathering their times (via
--timing and --verbose) and discarding slowest and fastest time (to get
rid of outliers):
- fast/css/should-not-insert-stylesheet-into-detached-document.html: 2.5s - 2.6s
- fast/dom/shadow/svg-style-in-shadow-tree-crash.html: 1.9s - 2.5s
- fast/encoding/char-encoding.html: 1.6s - 2.2s
- fast/frames/sandboxed-iframe-navigation-targetlink.html: 1.7s - 2.2s
- fast/html/marquee-destroyed-without-removed-from-crash.html: 1.8s - 2.7s
- fast/js/regress/string-cons-tower.html: 1.7s - 2.5s
- imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html: 2s -
2.7s
- virtual/pointerevent/fast/events/tabindex-focus-blur-all.html: 2.7s - 3s

BUG=24182, 577381, 477150
==========

[Łukasz Anforowicz, 2016-02-04 19:48:55]

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

[Łukasz Anforowicz, 2016-02-04 19:48:56]

Daniel, could you please rubberstamp these test expectation changes (marking a
few tests as "slow" to increase their default timeouts)?

This should be a safe change, as it doesn't really change the expectation that
the test passes, it just gives it a bit longer timeout.

FWIW, I've manually checked that these tests are not present in any other
expectation files (except
fast/html/marquee-destroyed-without-removed-from-crash.html which is also
present in LeakExpectations and fast/events/tabindex-focus-blur-all.html which
is marked as Timeout in FlagExpectations/enable-browser-side-navigation, but
then lint-test-expectations is happy so this should be fine I think).

[Łukasz Anforowicz, 2016-02-04 19:52:15]

carlosk@ to CC - I just wanted to bring to your attention that in this CL I am
marking fast/events/tabindex-focus-blur-all.html as slow in
third_party/WebKit/LayoutTests/SlowTests.  I've noticed that this test is marked
as [Timeout] in FlagExpectations/enable-browser-side-navigation, so maybe you
could try marking it as Slow instead (i.e. maybe it would pass even with
PlzNavigate, but just needs to take a bit longer).

[dcheng, 2016-02-04 20:41:34]

lgtm, but do we know why these tests are slow? Are they just doing a lot of
stuff? Are they polling for updates? Something else?

[Łukasz Anforowicz, 2016-02-04 21:45:33]

On 2016/02/04 20:41:34, dcheng wrote:
> lgtm, but do we know why these tests are slow? Are they just doing a lot of
> stuff? Are they polling for updates? Something else?

They do relatively lot of stuff (all of them either do more than 1 iteration or
test more than 1 input).  Some of them have sleeps / setTimeout (one at 100ms,
two for a relatively small 10ms).

fast/css/should-not-insert-stylesheet-into-detached-document.html
- 10 x (setTimeout(..., 100ms) + gc + reloading a page + some dom manipulation)

fast/dom/shadow/svg-style-in-shadow-tree-crash.html
- 10 x (setTimeout(..., 10ms) + svg-custom-html-element + css)

fast/encoding/char-encoding.html
- 37 x (submit-a-form + reload-a-document)

fast/frames/sandboxed-iframe-navigation-targetlink.html
- 11 x frame navigation

fast/html/marquee-destroyed-without-removed-from-crash.html
- 5 x (setTimeout(..., 10ms) + gc + reloading a page with frame with svg)

fast/js/regress/string-cons-tower.html
- 10000 x (100 x a = new String(a))

fast/events/tabindex-focus-blur-all.html
- 366 focus / blur events dispatched

imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html
- looks like multiple tests rolled into one?

[Łukasz Anforowicz, 2016-02-04 21:45:37]

The CQ bit was checked by lukasza@chromium.org

[commit-bot: I haz the power, 2016-02-04 21:47:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1668703006/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1668703006/1

[commit-bot: I haz the power, 2016-02-04 22:59:31]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-02-04 23:00:55]

Description was changed from

==========
Marking a few extra layout tests as slow.

When running tryjobs against crrev.com/1656033004 and
crrev.com/1660653005 I noticed flaky timeouts on linux_blink_dbg that
didn't seem related to my changes.  Running the tests locally revealed
that they are indeed a bit slow and that we should mark them as such (to
reduce tryjob flakiness).

Below results are from my local machine (Z420, Linux, Debug build) after
running tests 10 times (via --iterations=10), gathering their times (via
--timing and --verbose) and discarding slowest and fastest time (to get
rid of outliers):
- fast/css/should-not-insert-stylesheet-into-detached-document.html: 2.5s - 2.6s
- fast/dom/shadow/svg-style-in-shadow-tree-crash.html: 1.9s - 2.5s
- fast/encoding/char-encoding.html: 1.6s - 2.2s
- fast/frames/sandboxed-iframe-navigation-targetlink.html: 1.7s - 2.2s
- fast/html/marquee-destroyed-without-removed-from-crash.html: 1.8s - 2.7s
- fast/js/regress/string-cons-tower.html: 1.7s - 2.5s
- imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html: 2s -
2.7s
- virtual/pointerevent/fast/events/tabindex-focus-blur-all.html: 2.7s - 3s

BUG=24182, 577381, 477150
==========

to

==========
Marking a few extra layout tests as slow.

When running tryjobs against crrev.com/1656033004 and
crrev.com/1660653005 I noticed flaky timeouts on linux_blink_dbg that
didn't seem related to my changes.  Running the tests locally revealed
that they are indeed a bit slow and that we should mark them as such (to
reduce tryjob flakiness).

Below results are from my local machine (Z420, Linux, Debug build) after
running tests 10 times (via --iterations=10), gathering their times (via
--timing and --verbose) and discarding slowest and fastest time (to get
rid of outliers):
- fast/css/should-not-insert-stylesheet-into-detached-document.html: 2.5s - 2.6s
- fast/dom/shadow/svg-style-in-shadow-tree-crash.html: 1.9s - 2.5s
- fast/encoding/char-encoding.html: 1.6s - 2.2s
- fast/frames/sandboxed-iframe-navigation-targetlink.html: 1.7s - 2.2s
- fast/html/marquee-destroyed-without-removed-from-crash.html: 1.8s - 2.7s
- fast/js/regress/string-cons-tower.html: 1.7s - 2.5s
- imported/web-platform-tests/html/syntax/parsing/html5lib_tests21.html: 2s -
2.7s
- virtual/pointerevent/fast/events/tabindex-focus-blur-all.html: 2.7s - 3s

BUG=24182, 577381, 477150

Committed: https://crrev.com/8af8c1542f97215f2acad97cc5dbccfa81889c89
Cr-Commit-Position: refs/heads/master@{#373647}
==========

[commit-bot: I haz the power, 2016-02-04 23:00:56]

Patchset 1 (id:??) landed as
https://crrev.com/8af8c1542f97215f2acad97cc5dbccfa81889c89
Cr-Commit-Position: refs/heads/master@{#373647}

[carlosk, 2016-02-05 12:29:09]

On 2016/02/04 19:52:15, Łukasz Anforowicz wrote:
> carlosk@ to CC - I just wanted to bring to your attention that in this CL I am
> marking fast/events/tabindex-focus-blur-all.html as slow in
> third_party/WebKit/LayoutTests/SlowTests.  I've noticed that this test is
marked
> as [Timeout] in FlagExpectations/enable-browser-side-navigation, so maybe you
> could try marking it as Slow instead (i.e. maybe it would pass even with
> PlzNavigate, but just needs to take a bit longer).

Thanks for copying me Łukasz. I'll test it locally and update our expectations
if that fixes it.