Some more information... Don't know how I missed this earlier, but it turns out the script defer isn't the issue - the bug occurs even with window.onload. Stupid me.
Anyway, here's a link that may explain the whole thing: http://blog.metawrap.com/blog/MicrosoftsEolasActiveXPatchForIESometimesBreak sJavaScriptWorkaround.aspx -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tom Lee Sent: Wednesday, June 14, 2006 10:21 PM To: 'Flashcoders mailing list' Subject: RE: [Flashcoders] Eolas "fix" and backspace key flash bug Some good thoughts, Kevin. It'll take me some time to work through all the testing. I thought you'd be interested to know, I found two PCs the same version of IE on them - one exhibits the buggy behavior, the other doesn't. According to the version info, the IE installations are identical. However, there are some differences as to which Advanced Options are available, and one has the Java runtime installed and the other doesn't. Hopefully, I can narrow this down to a setting or something. On a very coarse-grain level, the installation that does not exhibit the bug is a much newer installation with no add-ons. I'm frankly not sure at this point whether to try to figure out the critical difference between the two browsers, or to try different JS methodologies until I hit one that works bug free. -tom -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Kevin Newman Sent: Wednesday, June 14, 2006 8:42 PM To: Flashcoders mailing list Subject: Re: [Flashcoders] Eolas "fix" and backspace key flash bug Maybe the workaround is to remove the old node, before overwriting it? I am currently just setting oldObjectNode.outerHTML = newObject.outerHTML. We might have to remove the old node completely, then reinsert it (in the correct position in the DOM). Actually, it's probably worth testing if the script is running at all when you set up your error condition (and whether the script has access to the DOM if it does run - an alert(document.body.innerHTML); should do). I think that you may have found a critical bug in the deferred script technique. I have noticed intermittent strange behavior with using deffered scripts with IE on other projects. Another solution might be to just suffer the extra load time (waiting for images and such), and use the document.onactivate event or document.onbeforeactivate, which fires before onload. I think right before it, which means after all the images load - actually, I'm not even sure it waits until the DOM is ready... If my assumptions about when this fires are correct, it will take longer to load than the defer method, but will still happen before onload, so onload scripting should still be safe. http://msdn.microsoft.com/workshop/author/dhtml/reference/events/onactivate. asp Then there's the old ondocumentready trick.. This is still related to Flash in a round about way, but it more about JScript ActiveX and IE than it is about Flash at this point. I'll let you guys decide if this should continue on the list... Kevin N. Tom Lee wrote: > Really I wasn't considering the delay as a real solution. It was just > something I was using to probe for clues. > > "the embedded object cannot exist prior to the > onload event firing or it will require activation. Period." > > Actually, that's not true. The only time it DOES require activation when > inserted by a deferred script (<script defer src="whatever.js"></script>) is > if the user clears their cache and revisits the page without restarting > their browser first. It is this edge case for which I am seeking a > workaround. > > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of ryanm > Sent: Wednesday, June 14, 2006 5:20 PM > To: Flashcoders mailing list > Subject: Re: [Flashcoders] Eolas "fix" and backspace key flash bug > > >> Perhaps there is a magic sweet spot between the execution >> of deferred scripts and window.onload when elements can be inserted into >> the >> document without "click to activate". >> >> > There's no magic there, the embedded object cannot exist prior to the > onload event firing or it will require activation. Period. Don't use a > delay, use the onload event. A delay is unpredictable because you are > working on the client side, where bandwidth, network congestion, and the > size of the images and other files loaded into the page will cause the > amount of time before the onload event fires to vary drastically. To state > that more clearly, you will *never* accomplish this in any predictable way > using a time delay, not even if you make the delay over 1 minute, because > somewhere there is still a guy using a 14.4 modem. > > ryanm > _______________________________________________ [email protected] To change your subscription options or search the archive: http://chattyfig.figleaf.com/mailman/listinfo/flashcoders Brought to you by Fig Leaf Software Premier Authorized Adobe Consulting and Training http://www.figleaf.com http://training.figleaf.com _______________________________________________ [email protected] To change your subscription options or search the archive: http://chattyfig.figleaf.com/mailman/listinfo/flashcoders Brought to you by Fig Leaf Software Premier Authorized Adobe Consulting and Training http://www.figleaf.com http://training.figleaf.com _______________________________________________ [email protected] To change your subscription options or search the archive: http://chattyfig.figleaf.com/mailman/listinfo/flashcoders Brought to you by Fig Leaf Software Premier Authorized Adobe Consulting and Training http://www.figleaf.com http://training.figleaf.com
