Both "leak" but not in the classic sense. IE uses 2 separate instances of
the DOM. One is DOM in javascript, One is the DOM of the HTML that's
rendered in the browser. This leak we are seeing is not orphaned tags
(Which is what programs like sieve keep track of). The leak is javascript
DOM memory that's being used but not reclaimed once the DOM elements are
removed. Reason being, in certain cases IE will crash if a code block tries
to reference DOM elements that no longer exists. In general using
removeChild in IE causes the browser to do a poor job identifying what
references can and cannot be garbage collected. These inner/outer HTML
methods seem to force IE's hand into doing a better (although not great) job
of cleaning up.
After running the tests at 2000 iterations it seems as though the outer
method is just about as effcient as cleaning up as the innerHTML method.
Essentially it comes down to changing the remove method to be the following:
remove: function( selector ) {
if ( !selector || jQuery.filter( selector, [ this ] ).r.length ) {
// Prevent memory leaks
jQuery( "*", this ).add(this).each(function(){
jQuery.event.remove(this);
jQuery.removeData(this);
});
if (this.parentNode) {
if (jQuery.browser.msie) {
this.outerHTML = '';
}
else {
this.parentNode.removeChild(this);
}
}
}
}
the "evalScript" and "clean" methods also use removeChild under the hood but
im not so sure they get used often enough to warrent a change.
-Chris
On Wed, Dec 3, 2008 at 8:22 AM, Ariel Flesler <[EMAIL PROTECTED]> wrote:
>
> So... let me get this right, both leak but innerHTML leaks less ?
>
> On Wed, Dec 3, 2008 at 2:45 AM, Chris Robinson <[EMAIL PROTECTED]> wrote:
> >
> > After a good deal of testing at 250 test runs it seems that the innerHTML
> > "garbage bin" and outerHTML methods are almost identical when it comes to
> > memory usage. outerHTML uses a VERY small amount more (around 36k). I
> am
> > going to rerun the test at 2000 iterations to see if it makes a
> difference.
> >
> > Again, sorry for the delay, work is very hectic these days. I should be
> > able to report back tomorrow during the day (fingers crossed).
> >
> >
> > On Mon, Dec 1, 2008 at 3:26 PM, Andrea Giammarchi
> > <[EMAIL PROTECTED]> wrote:
> >>
> >> Just as extra info, it seems that innerHTML with IE does a good work, it
> >> is like minimizing the window and maximizing it back. The problem is
> that it
> >> is not usable to force the CollectGarbage (that does not do its work as
> >> expected to me ...) ... does anybody else has bad experience with IE and
> >> ExtJS over jQuery? :D
> >>
> >> On Thu, Nov 27, 2008 at 7:12 PM, Andrea Giammarchi
> >> <[EMAIL PROTECTED]> wrote:
> >>>
> >>> Ariel I did a quick test with I don't remember which document and I
> >>> received an error.
> >>> Unfortunately I had no time to investigate but I have exactly the same
> >>> problem, "empty does not empty" at all.
> >>>
> >>> If the trick with outerHTML works I'll try tomorrow in my application
> and
> >>> I'll tell you.
> >>>
> >>> Regards
> >>>
> >>> On Thu, Nov 27, 2008 at 1:39 PM, Ariel Flesler <[EMAIL PROTECTED]>
> >>> wrote:
> >>>>
> >>>> So how do you know some cause errors if you haven't tested yet ?
> >>>>
> >>>> On Thu, Nov 27, 2008 at 9:47 AM, Andrea Giammarchi
> >>>> <[EMAIL PROTECTED]> wrote:
> >>>> > I guess textNodes, html itself and probably iframes.
> >>>> > I'll tell you as soon as I can test stuff :-)
> >>>> >
> >>>> > On Wed, Nov 26, 2008 at 1:21 PM, Ariel Flesler <[EMAIL PROTECTED]>
> >>>> > wrote:
> >>>> >>
> >>>> >> Which nodes generate an error ?
> >>>> >>
> >>>> >> On Wed, Nov 26, 2008 at 5:53 AM, Andrea Giammarchi
> >>>> >> <[EMAIL PROTECTED]> wrote:
> >>>> >> > Hi Ariel, is this what you would apply for each nested node in
> case
> >>>> >> > of
> >>>> >> > $.empty() ?
> >>>> >> >
> >>>> >> > The only problem I can see is with nodes that generate an error
> if
> >>>> >> > outerHTML
> >>>> >> > property is assigned and a try catch could decrease execution
> >>>> >> > speed.
> >>>> >> >
> >>>> >> > Any ideas?
> >>>> >> >
> >>>> >> > Regards
> >>>> >> >
> >>>> >> > On Tue, Nov 25, 2008 at 11:30 PM, Ariel Flesler
> >>>> >> > <[EMAIL PROTECTED]>
> >>>> >> > wrote:
> >>>> >> >>
> >>>> >> >> Well... I concluded a nicer solution and asked you to try it.
> >>>> >> >> But you never replied...
> >>>> >> >>
> >>>> >> >> http://dev.jquery.com/ticket/3553
> >>>> >> >>
> >>>> >> >> if (this.parentNode)
> >>>> >> >> this.parentNode.removeChild( this );
> >>>> >> >> if (jQuery.browser.msie)
> >>>> >> >> this.outerHTML = "";
> >>>> >> >>
> >>>> >> >> --
> >>>> >> >> Ariel Flesler
> >>>> >> >> http://flesler.blogspot.com/
> >>>> >> >>
> >>>> >> >> On Nov 25, 3:28 pm, "Chris Robinson" <[EMAIL PROTECTED]>
> wrote:
> >>>> >> >> > Hey Nick,
> >>>> >> >> >
> >>>> >> >> > this was never completely resolved. The edited jQuery source
> I
> >>>> >> >> > was
> >>>> >> >> > playing
> >>>> >> >> > around with is
> >>>> >> >> >
> >>>> >> >> >
> >>>> >> >> > here:
> http://www.outsidethediv.com/memory-leak-test/js/jquery-1.2.6-updated.js
> >>>> >> >> >
> >>>> >> >> > You can try swapping your jquery include with it and see if
> that
> >>>> >> >> > helps
> >>>> >> >> > at
> >>>> >> >> > all. It seemed to help out a good deal in my case but was
> >>>> >> >> > rather
> >>>> >> >> > inconclusive in smaller stripped down tests.
> >>>> >> >> >
> >>>> >> >> > On Tue, Nov 25, 2008 at 9:30 AM, Nick <[EMAIL PROTECTED]>
> >>>> >> >> > wrote:
> >>>> >> >> >
> >>>> >> >> > > I am having a very similar problem. Can someone give more
> >>>> >> >> > > details
> >>>> >> >> > > as
> >>>> >> >> > > how to modify the JQ source file to fix this?
> >>>> >> >> >
> >>>> >> >> > > I have noticed (in Drip) that when I refresh the page, the
> DOM
> >>>> >> >> > > nodes
> >>>> >> >> > > duplicate. This doesn't make any sense to me since I know
> JQ
> >>>> >> >> > > unbinds
> >>>> >> >> > > everything on page unload in IE. This node duplication is
> >>>> >> >> > > leading
> >>>> >> >> > > to
> >>>> >> >> > > some serious memory issues in my app that needs to run for
> up
> >>>> >> >> > > to 8
> >>>> >> >> > > hours straight without a browser restart.
> >>>> >> >> >
> >>>> >> >> > > Any help would be much appreciated.
> >>>> >> >> >
> >>>> >> >> > > Thanks,
> >>>> >> >> > > Nick
> >>>> >> >> >
> >>>> >> >> > > On Oct 24, 12:50 pm, chris robinson <[EMAIL PROTECTED]>
> >>>> >> >> > > wrote:
> >>>> >> >> > > > Hey all,
> >>>> >> >> >
> >>>> >> >> > > > I believe I have found a leak and have the beginnings of a
> >>>> >> >> > > > solution
> >>>> >> >> > > > to
> >>>> >> >> > > > it.
> >>>> >> >> >
> >>>> >> >> > > > I'm writing an app that preforms ajax searching that
> returns
> >>>> >> >> > > > pretty
> >>>> >> >> > > > large result sets. I render them via jQuery into a tbody.
> >>>> >> >> > > > once
> >>>> >> >> > > > a
> >>>> >> >> > > > subsequent search is performed I call tbody.empty() and
> >>>> >> >> > > > append
> >>>> >> >> > > > the
> >>>> >> >> > > > new
> >>>> >> >> > > > results. I created a test harness that would perform a
> >>>> >> >> > > > predefined
> >>>> >> >> > > > search 5000 times in a row with a good amount of time in
> >>>> >> >> > > > between.
> >>>> >> >> >
> >>>> >> >> > > > I noticed that IE was allocating quite a bit of memory and
> >>>> >> >> > > > never
> >>>> >> >> > > > reclaiming it. At first I thought it was a closure or
> >>>> >> >> > > > circular
> >>>> >> >> > > > reference on my part. Once I was sure I had removed them
> I
> >>>> >> >> > > > ran
> >>>> >> >> > > > more
> >>>> >> >> > > > test, sure enough it was still allocating a lot of memory.
> >>>> >> >> > > > Through
> >>>> >> >> > > > a
> >>>> >> >> > > > lot of research I found this articlehttp://
> >>>> >> >> > >www.scribd.com/doc/2159768/Ajax-Part2
> >>>> >> >> > > > which says that JS's removeChild will leak in IE.
> Microsoft
> >>>> >> >> > > > uses
> >>>> >> >> > > > another method to remove which is essentially this:
> >>>> >> >> >
> >>>> >> >> > > > function DestroyElement(elem) {
> >>>> >> >> > > > var garbageBin =
> >>>> >> >> > > > document.getElementById('IEMemoryLeakGarbageBin');
> >>>> >> >> > > > if(garbageBin === undefined) {
> >>>> >> >> > > > garbageBin = document.createElement("DIV");
> >>>> >> >> > > > garbageBin.id = "IEMemoryLeakGarbageBin";
> >>>> >> >> > > > garbageBin.style.display = 'none';
> >>>> >> >> > > > document.body.appendChild(garbageBin);
> >>>> >> >> > > > }
> >>>> >> >> > > > garbageBin.appendChild(elem);
> >>>> >> >> > > > garbageBin.innerHTML = "";
> >>>> >> >> >
> >>>> >> >> > > > }
> >>>> >> >> >
> >>>> >> >> > > > I went through jQuery 1.2.6 and replaced the removeChild
> >>>> >> >> > > > references
> >>>> >> >> > > > with a check for IE, and if so use this, else use the
> >>>> >> >> > > > regular
> >>>> >> >> > > > removeChild. After doing so and rerunning my test I saw
> >>>> >> >> > > > drastic
> >>>> >> >> > > > improvements in memory being reallocated after my elements
> >>>> >> >> > > > were
> >>>> >> >> > > > removed from the document.
> >>>> >> >> >
> >>>> >> >> > > > I also ran these test on FF, with or without this change
> it
> >>>> >> >> > > > ran
> >>>> >> >> > > > the
> >>>> >> >> > > > same way, recollecting memory correctly.
> >>>> >> >> >
> >>>> >> >> > > > this is only seems to be a drastic performance increase if
> >>>> >> >> > > > you
> >>>> >> >> > > > are
> >>>> >> >> > > > creating 1000+ dom elements and binding events to them,
> but,
> >>>> >> >> > > > the
> >>>> >> >> > > > app
> >>>> >> >> > > > I
> >>>> >> >> > > > am writing has to be able to run all day with out leaving
> or
> >>>> >> >> > > > refreshing the page.
> >>>> >> >> >
> >>>> >> >> > > > I just thought the Dev team might be interested in my
> >>>> >> >> > > > findings.
> >>>> >> >> > > > I
> >>>> >> >> > > > plan on striping this down and writing conclusive tests
> and
> >>>> >> >> > > > documentation over the weekend.
> >>>> >> >> >
> >>>> >> >> > > > -Chris
> >>>> >> >> >
> >>>> >> >> >
> >>>> >> >>
> >>>> >> >
> >>>> >> >
> >>>> >> > >
> >>>> >> >
> >>>> >>
> >>>> >>
> >>>> >>
> >>>> >> --
> >>>> >> Ariel Flesler
> >>>> >> http://flesler.blogspot.com
> >>>> >>
> >>>> >>
> >>>> >
> >>>> >
> >>>> > >
> >>>> >
> >>>>
> >>>>
> >>>>
> >>>> --
> >>>> Ariel Flesler
> >>>> http://flesler.blogspot.com
> >>>>
> >>>>
> >>>
> >>
> >>
> >>
> >
> >
> > >
> >
>
>
>
> --
> Ariel Flesler
> http://flesler.blogspot.com
>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"jQuery Development" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/jquery-dev?hl=en
-~----------~----~----~----~------~----~------~--~---
