680090 - jquery-powered elements are jittery/stuttering and slow in FF6

Status: RESOLVED INCOMPLETE

(Firefox :: General, defect)

Description

[Emlyn]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.5; rv:6.0) Gecko/20100101 Firefox/6.0
Build ID: 20110811165603

Steps to reproduce:

jquery elements like sliders, mouseovers and other visual effects that were previously working smoothly in FF5.x are now jittery, stuttering and slow in FF6. Suddenly a very poor user experience.


Actual results:

jquery animation/movement and mouse responsiveness are seriously impaired.

An example is here: http://www.imaginesurf.com


Expected results:

It should all be working smoothly...as with FF5.

Comment 1

[(mostly gone) XtC4UaLL [:xtc4uall]]

At least on Win32 (XP) I fail to see a Difference amongst 5, 6 and Trunk Builds.

Care to find a Regression Range?
http://harthur.github.com/mozregression/

Comment 2

[Vlad [QA]]

WFM on Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:9.0a1) Gecko/20110829 Firefox/9.0a1

Also, could you see if the issue occurs if using Firefox in safe mode:
http://support.mozilla.com/kb/Safe+Mode

How about with a new, empty testing profile? (Don't install any addons into it)
http://support.mozilla.com/kb/Basic+Troubleshooting#w_8-make-a-new-profile

Comment 3

[Emlyn]

Safe mode: Same behavior; jagged animation/scrolling.

New profile: Unable to complete this--terminal error:

"dyld: unknown required load command 0x80000022
Trace/BPT trap"

Even beyond just jquery/javascript page responsiveness, I've noticed a MARKED decline in overall performance of this browser. I get frequent, persistent hanging/freezes (spinning beach ball) and generally poor responsiveness.

Will plan to revert to version 5.x.

Comment 4

[Emlyn]

I should add (to make things more complicated) that it's not a constant behavior; it goes in and out of jagged animations/spinning beach balls. Simpler mouseovers and window behavior seems basically OK but sliding elements (ie: jquery >animate< functions) seem less responsive/smooth but not 100% of the time.

Comment 5

[Markus Stange [:mstange]]

Emlyn: Are you on OS X, too?

Comment 6

[Markus Stange [:mstange]]

Oh, beach ball, so yes.

Comment 7

[Emlyn]

FYI: [url=https://bugzilla.mozilla.org/show_bug.cgi?id=669034]Bug 669034[/url] and [url=https://bugzilla.mozilla.org/show_bug.cgi?id=667250]Bug 667250[/url] seem to point to the Javascript engine as a potential culprit.

The "known bug" [url=https://bugzilla.mozilla.org/show_bug.cgi?id=579260]Bug 579260[/url] for version 6 also references JSON/Javascript as a potential problem for scrolling (as well as "large profiles"...not sure what this refers to).

This problem is most noticeable for me when pages are javascript-intensive, have lots of nested content (iframes), and usually when I have lots of tabs open and programs running. ie: Possibly something memory-related. (iMac with 3Gb RAM) I also notice that google maps -> street view gets seriously laggy, as does some text entry.

Comment 8

[Emlyn]

(Sorry about the formatting here...didn't realize bug titles were automatically linked)

Comment 9

[Randell Jesup [:jesup] (needinfo me)]

This might be related to issues with saving sessionrestore data - can you post what the size of sessionstore.js in your profile is?  A new profile would be a good test, but apparently you can't create one - please file a bug on that, and also that might be related to whatever is causing your problem.  (I can't see how, but without knowing what's causing the profile creation problem it's hard to speculate

See bug 669034 and related dependencies.

Comment 10

[Emlyn]

sessionstore.js is 1.8Mb

Not sure on the new profile issue...I don't know terminal syntax all that much and so the error doesn't mean much to me.

FYI: Today I had only a few tabs open and just one other program. I was filling out a textarea box on a very simple page form when the typing refresh suddenly stopped. Got a spinning beach ball for a few seconds and the missing type appeared... This is pretty typical behavior (only much worse) when I have a lot more stuff open and I'm surfing crazy.

Comment 11

[Randell Jesup [:jesup] (needinfo me)]

That behavior can be caused by several things; one of which is a very large sessionstore.  Others can be related to specific pages.  1.8MB is largish for a small number of tabs, but not extreme.  You aren't using any network shares or other possible IO blockers (encrypted FS?), especially for the profile directory?

If you can find out what the system thinks was going on (some type of system monitor that can watch the process, and see if the CPU use by Firefox was pegged or not during the pauses, or if memory use spiked, etc.

Comment 12

[Emlyn]

Yes, that makes sense. Will try to monitor CPU usage while browsing and post back here...

Comment 13

[Emlyn]

Update: I've found a potential culprit...and a good vehicle for testing this, which may help to illustrate a potential memory bug in Firefox (although don't quote me on that yet).

On Facebook there's an online game called "Words with Friends" that is Flash-based...at least, if I had to guess. After this game has been open for some time (hours), the browser becomes progressively more sluggish. My guess is this is just poor code that, for whatever reason, becomes a memory hog the longer the tab is left open. (A very good example of this is having half a dozen games going at once (still in a single tab) which seems to accelerate/exacerbate the problem.)

Since this discovery I tried testing for this behavior in Safari 5.0.6. After a few hours of activity, there was some intermittent slowness but nothing nearly as bad as with Firefox, which would hang/freeze with increasing regularity and eventually slow to a crawl.

Hopefully this is useful information for someone. I have 3Gb of RAM, iMac core 2 Duo @ 2GHz.

Comment 14

[Emlyn]

I've been monitoring my Mac's CPU/RAM usage via Activity Monitor and my belief is that there is a SERIOUS memory issue with Firefox 6.

With no other programs open and Firefox already using a healthy 1.2Gb of 3Gb of available RAM, just changing tabs (of 4 open) causes CPU usage to spike to 112%+ for a few seconds, all the while lagging far behind in responsiveness from the mouse click and ALWAYS accompanied by the spinning beach ball.

This is pretty ridiculous. It is now unusable so I'm moving to Safari.

Am I the only person who is reporting on this kind of sudden, massively incapacitating memory issue?

Comment 15

[Emlyn]

Edit: RAM usage is 322Mb not 1.2Gb...that was for "VSIZE".

Comment 16

[Emlyn]

UPDATE: Repairing permissions in OS X Disk Utility actually improved things moderately, although it's a mystery to me why a single update to a browser would cause all other processes/applications on the computer to become less responsive.

However, as with some of the other related bugs I've read here about FF 6, I'm still experiencing intermittent spinning beach balls during otherwise normal browsing, especially when JAVA applets or Flash apps are active.

Comment 17

[coutureguillaume]

I have the same issur(In reply to Emlyn from comment #4)
> I should add (to make things more complicated) that it's not a constant
> behavior; it goes in and out of jagged animations/spinning beach balls.
> Simpler mouseovers and window behavior seems basically OK but sliding
> elements (ie: jquery >animate< functions) seem less responsive/smooth but
> not 100% of the time.

I have the same issue with FF6 and FF7 on OSX. jQuery animations are stuttering and when I begin typing into a textarea, it freezes. I searched a lot to solve my problem cause it works very great on FF6,7 on XP, FF5 on osx, Chrome, Saf, IE8+, and so on.. 
I may have a clue for this bug. For me, it happens only into a div position fixed. I'm not willing to change my code cause it works really great everywhere else..

Comment 18

[Mike Taylor [:miketaylr]]

Hi Emlyn,

Does this problem still reproduce?

Comment 19

[Emlyn]

This was submitted 5 years ago...so I have no idea TBH. I moved to Chrome years ago :/

Comment 20

[Mike Taylor [:miketaylr]]

OK, thanks for following up. Lets close then given the original reported URL no longer exists.

(You should try Firefox again. ^_^)