780615 - Very janky panning on minitab.com tutorial page

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: Toolbar, defect)

Description

[Mark Finkle (:mfinkle) (use needinfo?)]

On my Galaxy Nexus, I found this page to have a horrible amount of jank while panning.
http://www.minitab.com/en-US/training/tutorials/default.aspx

It seems to "stick" on each header

Comment 1

[Mark Finkle (:mfinkle) (use needinfo?)]

William - Any chance you could try this page in Eideticker?

Comment 2

[Mark Finkle (:mfinkle) (use needinfo?)]

This might be a good page to profile

Comment 3

[William Lachance (:wlach)]

(In reply to Mark Finkle (:mfinkle) from comment #1)
> William - Any chance you could try this page in Eideticker?

Yup, I can look at this tomorrow or Wednesday.

Comment 4

[Mark Finkle (:mfinkle) (use needinfo?)]

Not sure if this is fx17 only. Waiting for regression window.

Comment 5

[Aaron Train [:aaronmt]]

Dont think this is a regression. I can reproduce this on mozilla-central as far back as mid April.

Comment 6

[William Lachance (:wlach)]

(In reply to William Lachance (:wlach) from comment #3)
> (In reply to Mark Finkle (:mfinkle) from comment #1)
> > William - Any chance you could try this page in Eideticker?
> 
> Yup, I can look at this tomorrow or Wednesday.

Hey, sorry I didn't actually get to this last week. Been swamped trying to get my Pandaboard working and then the eideticker dashboard decided to break on me on Friday.

I did give the site on a try on my phone though, and it doesn't seem very obviously janky to me. I did notice a bit, but it wasn't huge. To be honest I'm not sure if Eideticker will help very much in developing this, though I could certainly create a simple test based on this site that we can use to track improvements in frame rate, etc.

Comment 7

[Andreea Pod]

Good build: 2012-04-17
Bad build: 2012-04-18

Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=c61e7c3a232a&tochange=0c7e2911be75

Comment 8

[Aaron Train [:aaronmt]]

Can we reduce it further? There are over 110 changes-sets there.

Comment 9

[Benoit Girard (:BenWa)]

My guess would be these changesets, but a real regression search would be ideal:
(For changing touch events)
http://hg.mozilla.org/mozilla-central/rev/a43a33ddd2b3
(For changing panning code)
http://hg.mozilla.org/mozilla-central/rev/1588b18050f4
http://hg.mozilla.org/mozilla-central/rev/fb2fd2b621ba
http://hg.mozilla.org/mozilla-central/rev/15845084c84d
(For changing display items)
http://hg.mozilla.org/mozilla-central/rev/6dbdb799fa6d
http://hg.mozilla.org/mozilla-central/rev/9157e2b13913
(low priority guess in case the problem was never fully fixed)
http://hg.mozilla.org/mozilla-central/rev/6b917c007510

Comment 10

[Kartikaya Gupta (email:[e-mail])]

I grabbed a profile of the janky panning on this page. The profile can be viewed at http://people.mozilla.com/~bgirard/cleopatra/?report=6089765054157c243d50e466a038d2552c353056 - there are some symbols missing but it looks like painting background images is causing the jank.

Comment 11

[Kartikaya Gupta (email:[e-mail])]

Panning the page with paintflashing turned on also shows that the entire screen is getting repainted and uploaded when it really shouldn't be. Not sure why that's happening.

Comment 12

[Aaron Train [:aaronmt]]

Not sure if it's me, but this seems better as of (08/27) on my GN 4.1.1. Anyone else still see this?

Comment 13

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to Aaron Train [:aaronmt] from comment #12)
> Not sure if it's me, but this seems better as of (08/27) on my GN 4.1.1.
> Anyone else still see this?

I agree. This is much better than previous builds. But I still see the entire page flashing with paintflashing dev option turned on. Maybe it's just flashing less?

Comment 14

[Mark Finkle (:mfinkle) (use needinfo?)]

Much improved