Re: [flexcoders] Module Memory Leaks
Hi Guys, Would calling image.unloadAndStop() in a dispose routine for item rederers who load runtime images be equivalent to setting the image source to null and calling validateNow()? Thanks, Greg On Fri, Feb 12, 2010 at 5:16 PM, Alex Harui aha...@adobe.com wrote: Runtime-loaded images need to be “disposed()” Set the source to null and let it validate or call validateNow(). I’m not sure about embedded image assets. I’ll try to find time to look into that. We’ve found two remoteobject bugs recently. One is that, on every call, a few bytes leak. It isn’t that bad but if it causes fragmentation it can add up over time. The other is that if you instantiate a new RemoteObject and try to get rid of it, you can’t and it seems to leak a decent chunk of memory. I’m pretty sure the first bug will be fixed in 10.1. The other was only recently reported so I don’t know if it will make 10.1, and most folks I hope aren’t creating and destroying RemoteObjects very often. On 2/12/10 7:39 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Thanks Alex, I had gone around and around to get a certain module to unload and was unable to determine the cause. Your article was very informative and my suspicion is that an image loaded at runtime was at fault. But that really isn’t the same thing as compile-time embedded assets, is it? As far as remote objects are concerned – if they are invoked via some sort of Command class, that should avoid this problem, am I right? Jeff -- *From:* flexcoders@yahoogroups.com [mailto:flexcoders@yahoogroups.comflexcoders@yahoogroups.com] *On Behalf Of *Alex Harui *Sent:* Thursday, February 11, 2010 6:39 PM *To:* flexcoders@yahoogroups.com *Subject:* Re: [flexcoders] Module Memory Leaks In Flex 4, styles are “per-module” and thus much less likely to leak. That’s the only change. Everything else is the same. Are you seeing evidence that embedded image assets are leaking? On 2/11/10 4:24 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Getting modules to fully unload has been something I’ve spent (wasted?) a good deal of time on in the past and I’ve just finished reading with great interest, Alex’s article on the subject: http://blogs.adobe.com/aharui/2009/08/what_we_know_about_unloading_m.html My question is – how much of this is resolved in Flex 4? Because if most or all of the issues have been addressed, I’ll simply abandon my Flex 3 development of module-based applications and move to Flex 4 right now, especially given that Flex 4’s release is imminent (early spring?). Of particular concern is the bit in Alex’s article about unloading images and the fact that image assets could be loaded via Stylesheets, etc. Jeff Battershall Application Architect Dow Jones Indexes jeff.battersh...@dowjones.com (609) 520-5637 (p) (484) 477-9900 (c) -- Alex Harui Flex SDK Team Adobe System, Inc. http://blogs.adobe.com/aharui
Re: [flexcoders] Module Memory Leaks
I don’t know for sure, but I don’t think so. UnloadAndStop might only work for SWF content. The profiler will tell you the answer. On 3/10/10 1:12 PM, Greg Hess flexeff...@gmail.com wrote: Hi Guys, Would calling image.unloadAndStop() in a dispose routine for item rederers who load runtime images be equivalent to setting the image source to null and calling validateNow()? Thanks, Greg On Fri, Feb 12, 2010 at 5:16 PM, Alex Harui aha...@adobe.com wrote: Runtime-loaded images need to be “disposed()” Set the source to null and let it validate or call validateNow(). I’m not sure about embedded image assets. I’ll try to find time to look into that. We’ve found two remoteobject bugs recently. One is that, on every call, a few bytes leak. It isn’t that bad but if it causes fragmentation it can add up over time. The other is that if you instantiate a new RemoteObject and try to get rid of it, you can’t and it seems to leak a decent chunk of memory. I’m pretty sure the first bug will be fixed in 10.1. The other was only recently reported so I don’t know if it will make 10.1, and most folks I hope aren’t creating and destroying RemoteObjects very often. On 2/12/10 7:39 AM, Battershall, Jeff jeff.battersh...@dowjones.com http://jeff.battersh...@dowjones.com wrote: Thanks Alex, I had gone around and around to get a certain module to unload and was unable to determine the cause. Your article was very informative and my suspicion is that an image loaded at runtime was at fault. But that really isn’t the same thing as compile-time embedded assets, is it? As far as remote objects are concerned – if they are invoked via some sort of Command class, that should avoid this problem, am I right? Jeff From: flexcoders@yahoogroups.com http://flexcoders@yahoogroups.com [mailto:flexcod...@yahoogroups.com] On Behalf Of Alex Harui Sent: Thursday, February 11, 2010 6:39 PM To: flexcoders@yahoogroups.com http://flexcoders@yahoogroups.com Subject: Re: [flexcoders] Module Memory Leaks In Flex 4, styles are “per-module” and thus much less likely to leak. That’s the only change. Everything else is the same. Are you seeing evidence that embedded image assets are leaking? On 2/11/10 4:24 AM, Battershall, Jeff jeff.battersh...@dowjones.com http://jeff.battersh...@dowjones.com wrote: Getting modules to fully unload has been something I’ve spent (wasted?) a good deal of time on in the past and I’ve just finished reading with great interest, Alex’s article on the subject: http://blogs.adobe.com/aharui/2009/08/what_we_know_about_unloading_m.html My question is – how much of this is resolved in Flex 4? Because if most or all of the issues have been addressed, I’ll simply abandon my Flex 3 development of module-based applications and move to Flex 4 right now, especially given that Flex 4’s release is imminent (early spring?). Of particular concern is the bit in Alex’s article about unloading images and the fact that image assets could be loaded via Stylesheets, etc. Jeff Battershall Application Architect Dow Jones Indexes jeff.battersh...@dowjones.com http://jeff.battersh...@dowjones.com (609) 520-5637 (p) (484) 477-9900 (c) -- Alex Harui Flex SDK Team Adobe System, Inc. http://blogs.adobe.com/aharui
RE: [flexcoders] Module Memory Leaks
Thanks Alex, I had gone around and around to get a certain module to unload and was unable to determine the cause. Your article was very informative and my suspicion is that an image loaded at runtime was at fault. But that really isn't the same thing as compile-time embedded assets, is it? As far as remote objects are concerned - if they are invoked via some sort of Command class, that should avoid this problem, am I right? Jeff From: flexcoders@yahoogroups.com [mailto:flexcod...@yahoogroups.com] On Behalf Of Alex Harui Sent: Thursday, February 11, 2010 6:39 PM To: flexcoders@yahoogroups.com Subject: Re: [flexcoders] Module Memory Leaks In Flex 4, styles are per-module and thus much less likely to leak. That's the only change. Everything else is the same. Are you seeing evidence that embedded image assets are leaking? On 2/11/10 4:24 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Getting modules to fully unload has been something I've spent (wasted?) a good deal of time on in the past and I've just finished reading with great interest, Alex's article on the subject: http://blogs.adobe.com/aharui/2009/08/what_we_know_about_unloading_m.html My question is - how much of this is resolved in Flex 4? Because if most or all of the issues have been addressed, I'll simply abandon my Flex 3 development of module-based applications and move to Flex 4 right now, especially given that Flex 4's release is imminent (early spring?). Of particular concern is the bit in Alex's article about unloading images and the fact that image assets could be loaded via Stylesheets, etc. Jeff Battershall Application Architect Dow Jones Indexes jeff.battersh...@dowjones.com (609) 520-5637 (p) (484) 477-9900 (c) -- Alex Harui Flex SDK Team Adobe System, Inc. http://blogs.adobe.com/aharui
Re: [flexcoders] Module Memory Leaks
Runtime-loaded images need to be “disposed()” Set the source to null and let it validate or call validateNow(). I’m not sure about embedded image assets. I’ll try to find time to look into that. We’ve found two remoteobject bugs recently. One is that, on every call, a few bytes leak. It isn’t that bad but if it causes fragmentation it can add up over time. The other is that if you instantiate a new RemoteObject and try to get rid of it, you can’t and it seems to leak a decent chunk of memory. I’m pretty sure the first bug will be fixed in 10.1. The other was only recently reported so I don’t know if it will make 10.1, and most folks I hope aren’t creating and destroying RemoteObjects very often. On 2/12/10 7:39 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Thanks Alex, I had gone around and around to get a certain module to unload and was unable to determine the cause. Your article was very informative and my suspicion is that an image loaded at runtime was at fault. But that really isn’t the same thing as compile-time embedded assets, is it? As far as remote objects are concerned – if they are invoked via some sort of Command class, that should avoid this problem, am I right? Jeff From: flexcoders@yahoogroups.com [mailto:flexcod...@yahoogroups.com] On Behalf Of Alex Harui Sent: Thursday, February 11, 2010 6:39 PM To: flexcoders@yahoogroups.com Subject: Re: [flexcoders] Module Memory Leaks In Flex 4, styles are “per-module” and thus much less likely to leak. That’s the only change. Everything else is the same. Are you seeing evidence that embedded image assets are leaking? On 2/11/10 4:24 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Getting modules to fully unload has been something I’ve spent (wasted?) a good deal of time on in the past and I’ve just finished reading with great interest, Alex’s article on the subject: http://blogs.adobe.com/aharui/2009/08/what_we_know_about_unloading_m.html My question is – how much of this is resolved in Flex 4? Because if most or all of the issues have been addressed, I’ll simply abandon my Flex 3 development of module-based applications and move to Flex 4 right now, especially given that Flex 4’s release is imminent (early spring?). Of particular concern is the bit in Alex’s article about unloading images and the fact that image assets could be loaded via Stylesheets, etc. Jeff Battershall Application Architect Dow Jones Indexes jeff.battersh...@dowjones.com (609) 520-5637 (p) (484) 477-9900 (c) -- Alex Harui Flex SDK Team Adobe System, Inc. http://blogs.adobe.com/aharui
Re: [flexcoders] Module Memory Leaks
In Flex 4, styles are “per-module” and thus much less likely to leak. That’s the only change. Everything else is the same. Are you seeing evidence that embedded image assets are leaking? On 2/11/10 4:24 AM, Battershall, Jeff jeff.battersh...@dowjones.com wrote: Getting modules to fully unload has been something I’ve spent (wasted?) a good deal of time on in the past and I’ve just finished reading with great interest, Alex’s article on the subject: http://blogs.adobe.com/aharui/2009/08/what_we_know_about_unloading_m.html My question is – how much of this is resolved in Flex 4? Because if most or all of the issues have been addressed, I’ll simply abandon my Flex 3 development of module-based applications and move to Flex 4 right now, especially given that Flex 4’s release is imminent (early spring?). Of particular concern is the bit in Alex’s article about unloading images and the fact that image assets could be loaded via Stylesheets, etc. Jeff Battershall Application Architect Dow Jones Indexes jeff.battersh...@dowjones.com (609) 520-5637 (p) (484) 477-9900 (c) -- Alex Harui Flex SDK Team Adobe System, Inc. http://blogs.adobe.com/aharui
