FMS2 CRASHING! (almost seems to be random)

[BryanTheCrow]

I'm writing an application that utilizes the XMLSocket object, and the XML /
XMLNode objects. It's getting to the point where there is allot of code
building / parsing XML some very simple Messages. It seems that once I've
passed a certain threshhold of building/sending, then a-synchronously
recieving/parsing about 2 xml messages, FMS2 will crash when the application is
reloaded. It will work fine the first time the app instance is run after a
server restart, but if the app instance ends (via connections dropping, or via
application.shutdown()) the next time the app is run, there seems to be a 70%
chance that FMS2 will crash.

After looking at the core log file, I get this error every time FMS crashes:

2005-11-28 18:12:11 5060 (e)2581279 Assert failed in \server\js\jsXML.cpp line
1523 -

Sometimes this error only shows up once, but sometimes it shows up several
times in a row (a few seconds apart).

A look at the crash dump file lead me to this file path for the problematic
source file:
c:\flashfarm\depot\branches\FlashCom_Edison\main\t incan\server\js\jsXML.cpp


I've tried to narrow down in my code exactly what is invoking this, but since
it happens at random times throughout my code (I put traces all over to see
where it stops), I can't find it. It's also not really doing anything tricky
with XML right now, so that adds to the confusion. If I stop it after it only
recieves 1 XML response, then I don't ever seem to run into the issue, but once
I've sent / recieved 2 or more messages, the crash is inevitable, albiet it
will come at a random time. The one thing I can think of that I'm doing that
might be a little out of the ordinary is extending the XML and XMLNode objects
to give me an easy way to get a child node by name or name / attribute values.
It's very simple though. Here it is (the dots at the begnining of some lines
are used to make it more readable on this website... they're not actually in
the code):

// Extend the XML & XMLNode objects to return a childnode by name, or
undefined if the node doesn't exist.
// The second & third optional parameters will only return the node if the
named attribute / value
// combo are also found in the node.
XML.prototype.getChildNode = XMLNode.prototype.getChildNode =
function(sNodeName, sAttributeName, sAttributeValue) {
....if (this.hasChildNodes) {
........for(var x = 0; x < this.childNodes.length; x++) {
............if (this.childNodes[x].nodeName == sNodeName) {
................if (sAttributeName != null && sAttributeValue != null) {
....................if (this.childNodes[x].attributes[sAttributeName] ==
sAttributeValue) {
........................return this.childNodes[x];
....................}
................} else {
....................return this.childNodes[x];
................}
............}
........}
....}

....// The requested node was not found. Return undefined:
....return undefined;
}


Sorry for the long post, but this crash, from what I can tell, seems to be a
problem in FMS2, and not my code. If anyone knows how to get in touch with
Macromedia on bugs like this (I don't know if they read these boards), or if
anyone has any ideas on this or has run into the same things, please let me
know.

If someone from Macromedia could please look at the following cpp file on line
1523:
c:\flashfarm\depot\branches\FlashCom_Edison\main\t incan\server\js\jsXML.cpp

Please let me know what this might be. I'll understand if it's something that
can't be fixed until a later release, but if you could give me a hint as to
what I might be able to avoid using in my code in order to avoid this, I'd
greatly appreciate it!

Thanks,
-Bryan

[BryanTheCrow]

Sorry, forgot to note. The crash alert shows the following:

FMSCore.exe - Application Error

The instruction at "0x0051d504" referenced memory at "0x0074065". The memory
could not be "written".

Upon disasembling FMSCore.exe, I found the following line of assembly was
causing the error:
0051D504 mov dword ptr [esi+4],edi

I have the dump file if macromedia would like it to find out what method was
being invoked...

[JayCharles]

Just a note... if you want someone from macromedia to respond, this isn't the place to make it happen. Contact them directly or report a bug. I don't think I've ever seen anyone from MM post here.

[thatguy2g]

Quick update:

I posted about this issue on the Flashcomguru Forums and have been contacted
by Macromedia. I'll post the solution we come to on that site (for anyone
running into the same issue):

[url]http://www.flashcomguru.com/forum/forum_posts.asp?TID=1590[/url]

[BryanTheCrow]

Quick Update:

I posted about this issue on the Flashcomguru Forums and have been contacted
by Macromedia.

Once we reach a solution I'll post here, and on the Flashcomguru Forum:

[url]http://www.flashcomguru.com/forum/forum_posts.asp?TID=1590[/url]