I've already landed the changes discussed (tweaking the logic - I
landed the wrong patch, missed the '!') and making sure that there's a
try/catch.
http://github.com/jquery/jquery/commit/3f648c4e3abe236b8ec6a19822313be794e5a9df
I tested in Firefox, IE 6 & 7 with ActiveX disabled. IE 6 just stops
On Jan 5, 10:04 am, John Resig wrote:
> The current logic is imperfect (in that, theoretically, someone could
> be trying to use jQuery on a local file, in IE 7, with ActiveX
> disabled and it would error out) but for now that's something that I
> can live with.
Why live with it? It fails in IE6
The security bar appears whenever you use any script (JScript or VBScript
etc) in a web page run locally (My Computer zone), even if it is just a
simple 'alert(location.href)'. I think this only appears after the 'Eolas
patch' done in April 2006 (which required users to click to activate
ActiveX, u
On Jan 5, 10:04 am, John Resig wrote:
> The current logic is imperfect (in that, theoretically, someone could
> be trying to use jQuery on a local file, in IE 7, with ActiveX
> disabled and it would error out) but for now that's something that I
> can live with.
It will still fail in IE6 with Act
Sam is correct here - IE 7 fails to request local files using the
traditional method. I also agree with your note about trying to use
ActiveX, especially when it's disabled. Adapting his code a little bit
I was able to come up with a solution and land it:
http://github.com/jquery/jquery/commit/b228
Hi,
>From what I recall from a few years back, XMLHttpRequest asynchronous
requests do not work with local files, which I think may be the reason why
ActiveX is used. Maybe check if it is being used on a local page (not
tested), e.g.
xhr: function() {
if (location.protocol !== "file:" && window.
On Dec 31 2009, 1:52 pm, John Resig wrote:
> Landed and closed:http://dev.jquery.com/ticket/5735
John, good to see this change in jQuery.
Another ajax area that needs attention is the xhr creation:
// Create the request object; Microsoft failed to properly
// implement the XMLHttpRequest in IE7
> Do you have a test page that repeatedly makes ajax calls to exploit
> the memory leak issue?
Yes.
> A .7% increase in memory use doesn't raise any red flags to me, it
> could just be how IE handles the new code differently. You'd have to
> repeat the test many times and watch Drip to see if the
On Dec 31, 8:53 am, John Resig wrote:
> Looking at the new numbers I see:
> Without patch: 19763200
> With patch: 19910656
> So still about a 1/5MB increase in memory usage.
Do you have a test page that repeatedly makes ajax calls to exploit
the memory leak issue?
A .7% increase in memory use doe
Thanks for the suggestion Matt. Landed it here:
http://github.com/jquery/jquery/commit/6cb2945837ccca55204191a8e7a70b2b2486c28e
Looking at the new numbers I see:
Without patch: 19763200
With patch: 19910656
So still about a 1/5MB increase in memory usage.
--John
On Thu, Dec 31, 2009 at 8:53 A
On Dec 30, 10:58 pm, John Resig wrote:
> Interesting, I've, also, seen the = null; proposal before, but not the
> = function(){}; one. Doing some poking around I found mention of it
> here:http://www.ilinsky.com/articles/XMLHttpRequest/#bugs-ie-leak
I haven't inspected the jQuery ajax code much,
Interesting, I've, also, seen the = null; proposal before, but not the
= function(){}; one. Doing some poking around I found mention of it
here:
http://www.ilinsky.com/articles/XMLHttpRequest/#bugs-ie-leak
I just experimented around with the change and posted it into a separate branch:
http://gith
On Dec 30, 7:00 am, Julian Aubourg wrote:
> If only it could be enough to set onreadystatechange to null :(
I wasn't proposing to set onreadystatechange to null, I was proposing
to change it to |nop|, a no-op function (that is, a function that has
an empty body). Sorry, I should have been a littl
On Dec 30, 9:00 am, Julian Aubourg wrote:
> If only it could be enough to set onreadystatechange to null :(
The work-arounds for IE memory leaks have been well known for a long
time. There shouldn't be any need to resort to timers and polling.
Perhaps a good solution to the leaks wasn't immediat
If only it could be enough to set onreadystatechange to null :(
Anyway, just a note to let you all know the polling is syndicated between
all requests in the ajax rewriting (ie: only one timer no matter how many
requests are being made concurrently).
2009/12/30 jblebrun
> On Nov 13, 3:12 pm, Jo
On Nov 13, 3:12 pm, John Resig wrote:
> It's because Internet Explorer has serious memoryleakissues with
> using the traditional model. Perhaps we could use conditional comments
> in this case, since it's not something that we can feature detect.
Is there a reason that the repeatedly (in other ve
16 matches
Mail list logo
