Patrick, Thanks for such a quick response, that was some very good feedback which I appreciate!
In regard to the GC, I understand the concept, and know that these decisions were made to increase the performance of Flash Player 9. It's just a bummer that I can leave an app open all night with absolutely no automated or user interation and come back in the morning with the same amount of memory allocated to the firefox.exe process. (Much of it that had been allocated during these effects) In regard to the app I am building, it is an Intranet app that was meant as a web-based replacement for a vb app. It's an app that's meant to run 8 hours a day without being closed and has a lot of transitions between the views and need for modal alert windows. So for me, 10 mb a pop without release is a pretty big hit over the course of the day. At this point the application development has completed and it's been through both SIT and UAT. I really do appreciate you suggestion on extending the effect, that is definitely something I'll try if I run into this again on another project. (This one's out of budget ;-) I'm going to be speaking with Adobe support here in the next day or so to see if they have any tips on how to use the mx.controls.Alert.show() in a modal manner without a significant amount of new memory being allocated and not released. If they bring to light any new information on this topic, then I will make sure to post it to this thread. Thanks again for your feedback! -Jun --- In flexcoders@yahoogroups.com, Patrick Mineault <[EMAIL PROTECTED]> wrote: > > From what I understand, what Ted means is that the player asks for such > and such amount of memory when it requires some, and may or may not give > it back when it doesn't need it anymore. So the worst case scenario is > that the memory consumed by the player is the maximum amount of memory > since it was loaded (the ceiling). That's the worst case scenario > however. In my personal tests in Firefox 2 when the totalMemory goes > down the memory consumption in Firefox also goes down a few seconds > later. I am talking about an app here which can take a full 100MB of > memory in a few seconds then release it in a few more seconds later > (lots of BitmapData manipulation and Loaders). I think what Ted means is > that you can't expect the garbage collector to collect when you want it > to and you can't expect that when the garbage collector is run the > memory will go down immediately, although it should eventually based on > the player's and the browser's memory allocation heuristics. > > As for the effects though, I think it's possible that the instance of > the effect is still in memory after being used, but I doubt that's the > actual cause of the problem, since the effect doesn't really contain all > that much stuff in it. It's possible however that once the effect is > played it doesn't completely clean up after itself. Things to look out > for include filters, cacheAsBitmap set to true, wipe effect masks which > linger on, etc. For example, if you use have a tween from a blur filter > to set blurX from 4 to 0, you have to make sure after to set filters to > an empty array and set cacheAsBitmap to false (filters automatically set > cacheAsBitmap to true). You'd really have to look at the source code of > the actual effect you're using to see how it works internally and see if > it's doing the kind of cleanup you want it to do, and if not, then > perhaps you should try to extend the effect to do the garbage collection > yourself. > > Here's my advice however: don't bother doing this unless you're in the > final stages of the application development. 10 MBs lingering about is > not going to make your app crash. I would expect that at the end your > app will have more important memory bottlenecks than this one, like > heavy use of Repeaters, too much components initialized, etc. > > Patrick > > coderjun a écrit : > > > > Hello, > > > > I was wondering if anyone has run across memory consumption issues when > > using some of the eye candy in the Flex framework. > > > > Here are a couple of examples of what I am referring to: > > > > 1. The translucent overlay on the rest of the app when using Alert.show > > in modal mode. (vs. using the flag Alert.NONMODAL) > > > > 2. Using <mx:WipeUp/> and <mx:WipeDown/ > for the inEffect and outEffect > > of a component. > > > > I've been tracking the effect of using these effects by watching Task > > Manager|Mem Usage for the firefox.exe process right before and right > > after these effects run. We're talking 4-10MB of memory consumed each > > time these effects are run. (I know because I've done the process of > > tracking with the same code and the effects turned off resulting in > > 300KB-1.5MB) . > > > > My concern is that: > > > > 1. I can't programmatically (minus the unsupported tip given to us by > > gskinner > > http://www.gskinner .com/blog/ archives/ 2006/08/as3_ resource_ > > ma_2.html > > <http://www.gskinner.com/blog/archives/2006/08/as3_resource_ma_2.html>) > > force the Garbage Collector to run. > > > > 2. Since I can't force the Garbage Collector to run, the comments by Ted > > Patrick on Flash Player memory management and releasing memory back to > > the OS is highly concerning. > > (http://tech. groups.yahoo. com/group/ flexcoders/ message/48219 > > <http://tech.groups.yahoo.com/group/flexcoders/message/48219> | > > http://tech. groups.yahoo. com/group/ flexcoders/ message/48330 > > <http://tech.groups.yahoo.com/group/flexcoders/message/48330>) > > > > I understand that effects would consume memory to run, but I don't like > > thinking that once the memory has been allocated I have no control on if > > it will be given back to the Operating System. > > > > I've tried leaving my app open all night to see if the memory allocation > > would go down and it doesn't. (You would think since the effects were > > long done it would) Also, the app I'm working on is one that needs to > > run all day, so closing and reopening the browser is not an option. > > > > Has anyone noticed what I've noticed or have any insight on how to use > > effects without increasing the footprint so? > > > > -Jun > > > > >
