Jan Moesen Reporter
	Description • 9 years ago

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:47.0) Gecko/20100101 Firefox/47.0
Build ID: 20160205030204

Steps to reproduce:

A) CNN

1. Disable/uninstall Flash so the page below will use <video>
2. Open <http://money.cnn.com/2016/02/04/investing/oil-prices-space-us-inventories-supply-glut/index.html>
3. The video starts to play automatically
4. Click the video to pause it
5. The video pauses, but it also navigates to a new URL dedicated to the video, <http://money.cnn.com/video/news/economy/2014/12/18/the-story-behind-oils-plunge.cnnmoney?iid=EL>

B) Crowdfunding.gent

1. Open https://crowdfunding.gent/nl/projects/fietsherstelzuil
2. Click the purple "Doneer nu" button
3. Click on any of the form elements, e.g. the amount input or the "Verberg mijn naam" checkbox, or the pre-set amount buttons
4. The dialog box closes again, as if you had clicked outside of the dialog box (but the intended action, e.g. checking the checkbox, has happened, too)

Maybe some click event that keeps on bubbling because the default was not prevented? Either way, I think something has changed recently-ish: stable Firefox (44) works fine, but Developer Edition (46.0a2) and Nightly (47.0a1) do not.

	Jan Moesen Reporter
	Comment 1 • 9 years ago

Regression window from mozregression: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=691f2e687e46d1b166b88b0a3f9dbd480bff9afa

A quick search for "click", "event", "bubbl" and "dom" did not yield anything useful. Did I do something wrong?

The pushlog from before mozregression switched the bisection method to taskcluster:

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=4a559a618d6798eb9a8fdc559f5a7a00085e2062

Nothing that jumps out at me in there, either. Sorry.

	Loic
	Comment 2 • 9 years ago

I confirm the regression range with testcase "B) Crowdfunding.gent".
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=691f2e687e46d1b166b88b0a3f9dbd480bff9afa

	Alice0775 White
	Comment 3 • 9 years ago

(In reply to Loic from comment #2)
> 
> I confirm the regression range with testcase "B) Crowdfunding.gent".
> https://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=691f
> 2e687e46d1b166b88b0a3f9dbd480bff9afa

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=6c49bb716593014c07d3736ea2124f74bfe3bb06&tochange=691f2e687e46d1b166b88b0a3f9dbd480bff9afa

Suspect: f950b7a04741	Thinker K.F. Li — Bug 1226904 - Fix boundary checking for leaves collecting. r=roc

	Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)
	Comment 4 • 9 years ago

Tracking since this is a regression in 46. 
Matt or Daniel can you take a look since I think Thinker is on PTO right now?

	Daniel Holbert [:dholbert]
	Comment 5 • 9 years ago

Matt knows much more about the 3d-transforms stuff that thinker has been working on than I do (I've been filing bugs but I don't really know the code), so I'll let Matt reply to comment 4 if he can.

	Matt Woodrow (:mattwoodrow) Assignee
	Comment 6 • 9 years ago

Marking as a blocker of 1097464 since that's effectively where this all started.

	Matt Woodrow (:mattwoodrow) Assignee
	Comment 7 • 9 years ago

(In reply to Jan Moesen from comment #1)
> Regression window from mozregression:
> https://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=691f
> 2e687e46d1b166b88b0a3f9dbd480bff9afa
> 
> A quick search for "click", "event", "bubbl" and "dom" did not yield
> anything useful. Did I do something wrong?
> 
> The pushlog from before mozregression switched the bisection method to
> taskcluster:
> 
> https://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=35b211eaad1fa828064514c547057e4400e24459&tochange=4a55
> 9a618d6798eb9a8fdc559f5a7a00085e2062
> 
> Nothing that jumps out at me in there, either. Sorry.

Which testcase did you use for finding the regression range?


(In reply to Jan Moesen from comment #0)
 
> A) CNN

Are you sure this is a regression? I see the same behaviour in both Safari and Chrome.

	Matt Woodrow (:mattwoodrow) Assignee
	Comment 8 • 9 years ago

This fixes the second testcase.

I'll make an automated test for this over the weekend and request review.

	Loic
	Comment 9 • 9 years ago

Yes, the testcase A is wrong, it happens with FF44. The testcase B is right.

	Jan Moesen Reporter
	Comment 10 • 9 years ago

Sorry for the red herring. I got confused because I experienced both issues around the same time. (You should see me when *three* things happen at once.)

	Pulsebot
	Comment 13 • 9 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/a8b26015bfd9

	Gary [:streetwolf52]
	Comment 14 • 9 years ago

I'm running with this patch on inbound (Windows 10).  The example for CNN in the OP still works the same way.

	Jan Moesen Reporter
	Comment 15 • 9 years ago

> The example for CNN in the OP still works the same way.
Yeah, that was my mistake. See comment 7, comment 9 and comment 10.

	Carsten Book [:Tomcat]
	Comment 16 • 8 years ago
bugherder

https://hg.mozilla.org/mozilla-central/rev/a8b26015bfd9

	Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)
	Comment 17 • 8 years ago

Matt can you request uplift to aurora ? Thanks

	Matt Woodrow (:mattwoodrow) Assignee
	Comment 18 • 8 years ago

Comment on attachment 8719268 [details] [diff] [review]
Handle nested preserve-3d contexts

Approval Request Comment
[Feature/regressing bug #]: Bug 1226904 (preserve-3d rewrite)
[User impact if declined]: Incorrect hit testing on some 3d websites.
[Describe test coverage new/current, TreeHerder]: Manually test, mochitest added.
[Risks and why]: Low risk, change is specific to 3d code.
[String/UUID change made/needed]: None

	Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)
	Comment 19 • 8 years ago

Comment on attachment 8719268 [details] [diff] [review]
Handle nested preserve-3d contexts

OK in m-c, includes new test, we need this for APZ in 46.

	Carsten Book [:Tomcat]
	Comment 20 • 8 years ago
bugherder uplift

https://hg.mozilla.org/releases/mozilla-aurora/rev/d31634b88c5a