The Danosphere.

Recently while working on a project using Flash Media Server 2, I discovered a rather major issue with the server itself. If a user disconnects in a less-than-graceful manner, such as losing internet connection, or unplugging their LAN cable, or something of that nature, the server seems to think they are still connected! At first I didn’t believe it but after further testing, sure enough the connection were being left open in FMS2, and this led to all types of other problems when FMS2 tried to send data to clients that actually no longer exist!

I contacted Adobe about this issue, and they confirmed this was indeed a bug (of course I had to make a sample app to prove it to them first). While there is not immediate Adobe solution to this issue, with the help of some other FMS’ers I have come up with this temporary solution. It is basically a heartbeat on the server side which calls a function on each client and waits for a response. If no response is received within XX secs, it can be assumed the client is no longer connected. This solution works well except if you have an application that does a lot of data crunching. Because if the app is crunching data, it wont respond to the heartbeat until the data is processed, and this can lead to false disconnects, something to keep in mind.

Here is the file i sent to adobe, which both illustrates the issue, and has a workaround to detect the detected client faster than FMS2 does on its own.

Ghost Connection Workaround

This problem presented itself while workin on a project for class that required alot of dynamic loading, attachMovies, and so forth. So I’m attaching these movies but for some reason my movieclip.removeMovieClip() calls aren’t doing anything! This problem can be very frustrating as Macromedia / Adobe does little to nothing to document this. The issue is components and the depthManager.

If you have a component in the library (does not need to be on stage!) and try to use _root.getNextHighestDepth() you will get returned a depth in the reserved range. Then when you try to removeMovieClip() the command will fail silently because the clip is out of depth range. As far as I’ve seen this issue only applies to clips who get their depth from _root.getNextHighestDepth()… using getNextHighestDepth() on other clips seems to still work fine. Anyway here is an easy solution I found somewhere:

[actionscript]

function removemovieclip(removemovie:MovieClip) {

var scope = removemovie.parent;
var poorswappedmc = scope.getInstanceAtDepth(0);
poorswappedmc.swapDepths(scope.getNextHighestDepth());
removemovie.swapDepths(0);
removemovie.removeMovieClip();
poorswappedmc.swapDepths(0);

}

[/actionscript]

Simple really- swap the target mc back into the valid depth range and then remove! voila!

On yet another DL assignment I had to figure out how to get the browser to stop caching my swf’s. I tried using the Adobe/Macromedia Technote method of setting headers to force the refresh of content, but, big surprise, it didnt work.

So I searched for a different method. I needed the page to reload the swf regardless of if it was the latest or not, i wanted it to refresh every.single.time. The solution was simpler than I had expected. By simply appending an arbitrary random number variable onto the swf location the browser will force refresh the swf every time. Here is the code [using the IE activeX bypass]

[script language="javascript" type="text/javascript"]

var randNum = Math.round(100*Math.random());
AC_FL_RunContent(“src”, “myCachedSWF?cacheController=” + randNum);

[/script]