Simona Badau, Desktop QA Reporter
	Description • 8 years ago

Mozilla/5.0 (Windows NT 6.3; rv:48.0) Gecko/20100101 Firefox/48.0
Build ID: 20160314030215

[Affected versions]:
Nightly 48.0a1
Aurora 47.0a2

[Affected platforms]:
Windows 8.1
Mac OS X 10.11

[Steps to reproduce]:
1. Navigate to Google news: https://news.google.com/
2. Scroll down and then up again to reach the Google logo 

[Expected result]:
Scrolling is smooth.

[Actual result]:
Google Logo hangs when scrolling (please see the attached video for more details).

[Regression range]:
Please let me know if a regression range is needed.

[Additional notes]:
- the issue doesn't seems to be as noticeable on Ubuntu or on Firefox Beta 46.0b1 and Firefox 45 RC.

	Jamie Nicol [:jnicol]
	Comment 1 • 8 years ago

Thanks for the report. Since this is scrolling-related there's a chance this could be an APZ problem. A regression range could come in useful, but before you spend time finding one it might be worth checking if you can reproduce this with layers.async-pan-zoom.enabled = false in about:config (requires a restart after toggling).

Also, it would be useful to attach the about:support for Nightly on an affected platform and also on Ubuntu. (Here I'm wondering whether e10s and therefore APZ is disabled on Ubuntu for some reason, hence why it doesn't affect it.)

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 2 • 8 years ago

The video is pretty low framerate so it's hard to tell what the problem is, but I stepped through it in QuickTime and saw one frame where the logo was "overlapping" the left sidebar. Simona, can you confirm that this is the problem you're referring to?

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 3 • 8 years ago

The problem happens not because of the logo, but because the sidebar is changed from position:fixed to position:absolute, and so is delayed in repositioning properly. If they were to use position:sticky instead then it wouldn't happen. It's relatively minor compared to some of the other websites we've seen with this problem. This is a result of APZ, one of the scroll-linked effects.

	Simona Badau, Desktop QA Reporter
	Comment 4 • 8 years ago

(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #2)
> Created attachment 8731680 [details]
> Frame from video
> 
> The video is pretty low framerate so it's hard to tell what the problem is,
> but I stepped through it in QuickTime and saw one frame where the logo was
> "overlapping" the left sidebar. Simona, can you confirm that this is the
> problem you're referring to?

Yes, the "overlapping" is the problem I'm referring to.

	Botond Ballo [:botond]
	Comment 5 • 8 years ago

As per comment 3, the site should be using position:sticky. Moving to Tech Evangelism.

	Mike Taylor [:miketaylr]
	Comment 6 • 8 years ago

I get jank when I scroll in general on this site, but it's sort of intermittent -- only when I do sort of a fling up or down gesture with my mouse (if that makes any sense). Maybe every 4th fling.

Kats, can you reproduce the scroll jank?

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 7 • 8 years ago

Looking at the page source, it is still doing the same thing as before (switching between position:fixed and position:absolute) but I'm not actually able to reproduce the overlap - I think the switch happens quickly enough that it's not really noticeable. However if the main thread is stuck/slow for whatever reason then it would probably still be noticeable.

	Milan Sreckovic [:milan] (needinfo for best results)
	Comment 8 • 8 years ago

Can we confirm the status here?  Do we know it's evangelism issue, or could there be a bug on our side?  Or is it a fuzzy spec?

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 9 • 8 years ago

It's one of the fallout issues inherent in APZ. We would like them to use position:sticky instead of switching between position:fixed and position:absolute based on the scroll offset, hence evangelism.

	Mike Taylor [:miketaylr]
	Comment 10 • 8 years ago

Karl, can we ask Google to use position:sticky here? They're working on an implementation so it will be good for them at some point in the future as well.

	Karl Dubost💡 :karlcow Assignee
	Comment 11 • 8 years ago

I had hard time noticing it in the video too. 
And I can't reproduce on Nightly 51.0a1 (2016-08-17) on macosx. 
And I can't reproduce on Firefox 48.

I will contact Google, but I have the feeling that will not be part of their priorities.

Contacted today

	Karl Dubost💡 :karlcow Assignee
	Comment 12 • 8 years ago

We got a response with two axis:
1. Why should we do it?
2. We will look into it.

for the 1st one I gave reasons.

1. https://www.chromestatus.com/features/6190250464378880
  and https://bugs.chromium.org/p/chromium/issues/detail?id=231752#c49
  If  I understand it's ready to go more public.

2. working in Safari and Firefox
  http://caniuse.com/#feat=css-sticky

3. Under consideration for Edge
  https://developer.microsoft.com/en-us/microsoft-edge/platform/status/positionsticky

4. Google sends different codes to Firefox and Chrome already. (Though we want to avoid this as much as possible.)

	Karl Dubost💡 :karlcow Assignee
	Comment 13 • 8 years ago

Google's answer

> "position: sticky" seems useful and I have raised this to our frontend folks.  
> It turned out, they already knew about this feature.  However, I don't know 
> of the ETA as it is a matter of their priorities when they get to test this 
> on Firefox on various platforms and also on other browsers.

	Mike Taylor [:miketaylr]
	Comment 14 • 8 years ago

Thanks Karl. Hopefully Google's implementation of pos:sticky will come sooner than later and it will be an easier sell.

	Karl Dubost💡 :karlcow Assignee
	Comment 15 • 7 years ago

> The great news is that as of Chrome 56 (currently beta 
> as of December 2016, stable in Jan 2017) position: sticky 
> is now back in Chrome.
https://developers.google.com/web/updates/2016/12/position-sticky

	Mike Taylor [:miketaylr]
	Comment 16 • 6 years ago

There's a new design, and it always uses a pos:fixed header, rather than attempting to emulate sticky. I think we can close now.