Re: SWF11 components performance/stability
Hi Alexei, FYI - I've added a post about the browser problems resolving in here: http://opensourcewebconferencing.blogspot.ru/2013/03/how-to-debug-crashes-in-firefox.html FAQ also contains this link now. Best regards, Irina. On Thu, Mar 14, 2013 at 5:28 PM, Alexei Fedotov alexei.fedo...@gmail.comwrote: Irina, please, add the possible reason for flash crashes due to a driver, a third party soft or a browser cache to the faq 14.03.2013 13:32 пользователь Irina Arkhipets irina.arkhip...@gmail.com написал: Yes, and actually I hope it would help my client too :) Regarding the video acceleration - actually, I just saw crash reports similar with mine in Mozilla bug tracker, and the recommendation was to turn this off. Regards, Irina. On Thu, Mar 14, 2013 at 4:24 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Hopefully the 26th tryout is no fail :) Glad we could track this down to be an environment issue. It would be nice to know if the hardware acceleration has an effect on the broadcasting (encoding) at all. I am pretty sure it has an impact on viewing / decoding. But I am not so sure for broadcasting/encoding video. Sebastian 2013/3/14 Irina Arkhipets irina.arkhip...@gmail.com Seems like it works for me. At least, after the ffdshow update I was able to enter/exit the room 25 times without crash. I'll look for this a couple of days. And ask my client who claimed about the problem to do the same things and additionally try to turn the hardware acceleration off. Thank you, Irina. On Thu, Mar 14, 2013 at 3:28 PM, Irina Arkhipets irina.arkhip...@gmail.comwrote: Sebastian, It seems like you are correct and this depends on configuration. I've received a reply from Mozilla about my bug: ... According to the previous stack trace, it's not a Firefox bug neither a Flash one, but a FFDShow one. You are using version 1.1.3761.0 while the current one is 1.3.4500. Update FFDShow: http://ffdshow-tryout.sourceforge.net/download.php Then report back. ... You can track this here: https://bugzilla.mozilla.org/show_bug.cgi?id=850614 Will try their suggestion and inform you about the result. Best regards, Irina. On Thu, Mar 14, 2013 at 3:03 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: After all this sounds a lot like an issue with the webcam, microphone or driver. The moment when your are in the conference, in the device settings and click on the button start conference what basically happens is that you set your audio/video settings and Flash access your webcam. So basically this is an environment issues. It could be also a combination of a Flash Player + certain device issues. I bet the same issue happens for those users if they use any other Flash application that requires access to cam/mic. What kind of microphone and camera are you using ? Did you have maybe some driver that you can update or download for those devices for your OS ? Sebastian 2013/3/14 Maxim Solodovnik solomax...@gmail.com @Alexei unfortunately this would not help since we have no useful info after such crash :( On Thu, Mar 14, 2013 at 2:33 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Maxim, how difficult is to add a button for debug which would start/quit a conference in a loop to make the bug simpler to reproduce? 14.03.2013 10:56 пользователь Irina Arkhipets irina.arkhip...@gmail.com написал: Hi Sebastian, A). I use Win XP + FF 19.0 + flash 11.6.602.180. And Yulia have the same problems on W7 + GH + 11.6.602.171. B). Crash happens when I am entering the room after I click Start Conference in the Choose device screen. C). Plugin crashes in the following way: I get plugin-container.exe application error window with the possibility to click OK to terminate and click CANCEL to debug. If I click OK I see a grey screen with the Adobe Flash plugin crashed message and the Send report button. I need to reload the page to continue work. D). It seems like it's not possible to attach files when you write to mailing list, so I'll will send you a screen shot personally. I've also found a way to catch a report which is sent if I click the Send report button - will send it to you too. E). Nothing special in the server log and browser console My additional notes: - Crash is reproducible if I run FireFox in safe mode. - Crash report reads: Crash Reason EXCEPTION_ACCESS_VIOLATION_WRITE. I read in Internet that it can happen because of some video driver problems if hardware acceleration is turmed on, however I've just turned it off without improving - I've filed
Re: SWF11 components performance/stability
No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.comwrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559 and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone and switch to h264 According to the Google switching to h264 can cause higher CPU usage. Can this be fixed by changing H264Level? (currently H264Level.LEVEL_1_2 is used) Maybe you can us understand were performance/stability degradation caused by code changes? Or maybe it is because of Flex4.6.0? Can stability be improved by switching to Flex4.9.1? Thanks in advance for your help [1] http://markmail.org/message/2el63qarsjmbvfyy http://markmail.org/message/2el63qarsjmbvfyy-- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com
Re: SWF11 components performance/stability
Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.comwrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559 and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone and switch to h264 According to the Google switching to h264 can cause higher CPU usage. Can this be fixed by changing H264Level? (currently H264Level.LEVEL_1_2 is used) Maybe you can us understand were performance/stability degradation caused by code changes? Or maybe it is because of Flex4.6.0? Can stability be improved by switching to Flex4.9.1? Thanks in advance for your help [1] http://markmail.org/message/2el63qarsjmbvfyy http://markmail.org/message/2el63qarsjmbvfyy-- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian
Re: SWF11 components performance/stability
You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.comwrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559 and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone and switch to h264 According to the Google switching to h264 can cause higher CPU usage. Can this be fixed by changing H264Level? (currently H264Level.LEVEL_1_2 is used) Maybe you can us understand were performance/stability degradation caused by code changes? Or maybe it is because of Flex4.6.0? Can stability be improved by switching to Flex4.9.1? Thanks in advance for your help [1]
Re: SWF11 components performance/stability
@Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.com wrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559 and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone
Re: SWF11 components performance/stability
@Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.comwrote: @Max AS3 means action script Modern OpenLaszlo compiles into AS3, these files can be inspected and used instead of openlaszlo files. Then standard Adobe tools create swf bytecode from AS3. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:18 AM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: MXML has certainly advantages in the sense of wider acceptance and better development tools. But it really means to build something from scratch. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.com wrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea
Re: SWF11 components performance/stability
The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Max AS3 means action script Modern OpenLaszlo compiles into AS3, these files can be inspected and used instead of openlaszlo files. Then standard Adobe tools create swf bytecode from AS3. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:18 AM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: MXML has certainly advantages in the sense of wider acceptance and better development tools. But it really means to build something from scratch. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote before this issue has been further analyzed. Is there a given restriction that needs h264? Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.com wrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com
Re: SWF11 components performance/stability
Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you compile such pointers. You need to pass the parameter into the object to access its attributes, or pass the attribute inside of it. So a lot of the things we do when having some attribute in the canvas and pointing to them from various places in the app ... it simply doesn't work in MXML. So of course it depends on to what extend you plan to build a Flex client. Maybe just the SWF11 app converting to MXML is an option. But the entire SWF8 app. I don't think so. Just too complex. Sebastian 2013/3/13 Alexei Fedotov alexei.fedo...@gmail.com The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Max AS3 means action script Modern OpenLaszlo compiles into AS3, these files can be inspected and used instead of openlaszlo files. Then standard Adobe tools create swf bytecode from AS3. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:18 AM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: MXML has certainly advantages in the sense of wider acceptance and better development tools. But it really means to build something from scratch. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with h264 I'm going to build 1) SWF11 with h264 support removed 2) SWF11 with different h264 profile 3) SWF10 and compare but i'm not expecting any major differences :( On Wed, Mar 13, 2013 at 1:17 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: No there is no such conversion. You have to rewrite the LZX to MXML and AS3 files. What about the Video Codec? I don't think it makes sense to start a RC2 vote
Re: SWF11 components performance/stability
just for consideration, attached is the current AS3 that OpenLaszlo generates. https://cwiki.apache.org/confluence/download/attachments/27836308/lzswf9.tar.gz Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com I thought maybe it will be simple to convert swf10 folder of our sources + I would like someone else will handle it : On Wed, Mar 13, 2013 at 4:38 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you compile such pointers. You need to pass the parameter into the object to access its attributes, or pass the attribute inside of it. So a lot of the things we do when having some attribute in the canvas and pointing to them from various places in the app ... it simply doesn't work in MXML. So of course it depends on to what extend you plan to build a Flex client. Maybe just the SWF11 app converting to MXML is an option. But the entire SWF8 app. I don't think so. Just too complex. Sebastian 2013/3/13 Alexei Fedotov alexei.fedo...@gmail.com The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Max AS3 means action script Modern OpenLaszlo compiles into AS3, these files can be inspected and used instead of openlaszlo files. Then standard Adobe tools create swf bytecode from AS3. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:18 AM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: MXML has certainly advantages in the sense of wider acceptance and better development tools. But it really means to build something from scratch. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However... all just guessing at this point. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com Maybe we can add conversion task to the GSOC ideas? do we need it? I don't see any problems with
Re: SWF11 components performance/stability
that is definitely something not readable :( I can see the word static in the generated code :) this is global variables :))) On Wed, Mar 13, 2013 at 4:57 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: just for consideration, attached is the current AS3 that OpenLaszlo generates. https://cwiki.apache.org/confluence/download/attachments/27836308/lzswf9.tar.gz Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com I thought maybe it will be simple to convert swf10 folder of our sources + I would like someone else will handle it : On Wed, Mar 13, 2013 at 4:38 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you compile such pointers. You need to pass the parameter into the object to access its attributes, or pass the attribute inside of it. So a lot of the things we do when having some attribute in the canvas and pointing to them from various places in the app ... it simply doesn't work in MXML. So of course it depends on to what extend you plan to build a Flex client. Maybe just the SWF11 app converting to MXML is an option. But the entire SWF8 app. I don't think so. Just too complex. Sebastian 2013/3/13 Alexei Fedotov alexei.fedo...@gmail.com The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Max AS3 means action script Modern OpenLaszlo compiles into AS3, these files can be inspected and used instead of openlaszlo files. Then standard Adobe tools create swf bytecode from AS3. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:18 AM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: MXML has certainly advantages in the sense of wider acceptance and better development tools. But it really means to build something from scratch. Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com @Alexey we currently using openlaszlo for video (not sure what do you mean by AS3). Currently our version of video components is built using flex.4.6.1 while actual version available for download is flex 4.9.1. I was hoping we will have less bugs and better performance if we will use MXML directly for our code. But this is just a guess. On Wed, Mar 13, 2013 at 2:00 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: You can simply disable h264 and build to swf11. It is only about the codec not about SWF formats. I just compared and could not really find out if there are big differences. However it is definitly good to really compare the impacts on CPU and bandwidth usage when switching the Codec for video encoding. It seemed to me like encoding with h264 is using a second process (on OSX that process is called WindowsServer). It might be that h264 encoding can use some more system ressources. However...
Re: SWF11 components performance/stability
Hi All, I investigated flash plugin crashes today and would like to share some results. First of all, crash happens both with SWF10 and SWF 11. It seems like the problem is reproducible with H263, H264 1.2 and H264 5.1. Actually, I haven't noticed significant CPU usage degradation for H264 in compare with H263 - it always was about 50-70% when I enter OpenMeetings room. The same is true about the memory usage. I was able to get the crash report in Mozilla, however for now don't see any info which can help me there. Tested environment was: Windows XP + Mozilla FireFox 19.0 + flash player 11.5.502.146. Tested server was demo.dataved.ru, there were two users with videos (1 with big window and the other with normal) in the room + me with normal window. For now anybody can try different cases on http://demo.dataved.ru/openmeetings and get own expression. Below are the links which can be tested and my today results for each of them: - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf10.swfjunk=666 - SWF10 + H263. Crashed 3 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h263.swfjunk=666- SWF11 + H263. Crashed 0 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_1_2.swfjunk=666- SWF11 + H264 1.2. Crashed 2 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_5_1.swfjunk=666- SWF11 + H264 5/1/ Crashed 1 times from 10. Best regards, Irina. On Wed, Mar 13, 2013 at 5:02 PM, Maxim Solodovnik solomax...@gmail.comwrote: that is definitely something not readable :( I can see the word static in the generated code :) this is global variables :))) On Wed, Mar 13, 2013 at 4:57 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: just for consideration, attached is the current AS3 that OpenLaszlo generates. https://cwiki.apache.org/confluence/download/attachments/27836308/lzswf9.tar.gz Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com I thought maybe it will be simple to convert swf10 folder of our sources + I would like someone else will handle it : On Wed, Mar 13, 2013 at 4:38 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you compile such pointers. You need to pass the parameter into the object to access its attributes, or pass the attribute inside of it. So a lot of the things we do when having some attribute in the canvas and pointing to them from various places in the app ... it simply doesn't work in MXML. So of course it depends on to what extend you plan to build a Flex client. Maybe just the SWF11 app converting to MXML is an option. But the entire SWF8 app. I don't think so. Just too complex. Sebastian 2013/3/13 Alexei Fedotov alexei.fedo...@gmail.com The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected? @Sebastian I thought we use only view/label/dataset/connection in our swf10 code all these primitives should be available in MXML. On Wed, Mar 13, 2013 at 2:52 PM, Alexei Fedotov alexei.fedo...@gmail.com wrote: @Max AS3 means action script
Re: SWF11 components performance/stability
Great job, Irina! Thanks for digging this out. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 4:28 PM, Irina Arkhipets irina.arkhip...@gmail.com wrote: Hi All, I investigated flash plugin crashes today and would like to share some results. First of all, crash happens both with SWF10 and SWF 11. It seems like the problem is reproducible with H263, H264 1.2 and H264 5.1. Actually, I haven't noticed significant CPU usage degradation for H264 in compare with H263 - it always was about 50-70% when I enter OpenMeetings room. The same is true about the memory usage. I was able to get the crash report in Mozilla, however for now don't see any info which can help me there. Tested environment was: Windows XP + Mozilla FireFox 19.0 + flash player 11.5.502.146. Tested server was demo.dataved.ru, there were two users with videos (1 with big window and the other with normal) in the room + me with normal window. For now anybody can try different cases on http://demo.dataved.ru/openmeetings and get own expression. Below are the links which can be tested and my today results for each of them: - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf10.swfjunk=666 - SWF10 + H263. Crashed 3 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h263.swfjunk=666- SWF11 + H263. Crashed 0 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_1_2.swfjunk=666- SWF11 + H264 1.2. Crashed 2 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_5_1.swfjunk=666- SWF11 + H264 5/1/ Crashed 1 times from 10. Best regards, Irina. On Wed, Mar 13, 2013 at 5:02 PM, Maxim Solodovnik solomax...@gmail.comwrote: that is definitely something not readable :( I can see the word static in the generated code :) this is global variables :))) On Wed, Mar 13, 2013 at 4:57 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: just for consideration, attached is the current AS3 that OpenLaszlo generates. https://cwiki.apache.org/confluence/download/attachments/27836308/lzswf9.tar.gz Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com I thought maybe it will be simple to convert swf10 folder of our sources + I would like someone else will handle it : On Wed, Mar 13, 2013 at 4:38 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you compile such pointers. You need to pass the parameter into the object to access its attributes, or pass the attribute inside of it. So a lot of the things we do when having some attribute in the canvas and pointing to them from various places in the app ... it simply doesn't work in MXML. So of course it depends on to what extend you plan to build a Flex client. Maybe just the SWF11 app converting to MXML is an option. But the entire SWF8 app. I don't think so. Just too complex. Sebastian 2013/3/13 Alexei Fedotov alexei.fedo...@gmail.com The files appear in some temporary directory. I cannot remember where they are exactly. If you make an error when embedding action script into openlaszlo (should be something on action script level, xml markup would break the preprocessor itself), you get in the error message an exact pointer to the file where your error is put. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 11:58 AM, Maxim Solodovnik solomax...@gmail.com wrote: @Alexey we currently using flex 4.5.0, 4.6.1 to create bytecode can be inspected/used means autoconverted :) action script instead of openlaszlo is also good :) can you please tell me how this can be inspected?
Re: SWF11 components performance/stability
I think the issue has been reviewed. So we can be sure that the codec change in the codec is not an issue. Maybe we can debug the crashes a bit more before we start the vote. @Irina Thanks for the Tests! Can you please check those questions A) What OS and flash player Version and browser was that? B) When did the crash happen exactly? C) Did the Plugin crash in terms of: 1) showing a dialog that a script is slowing down? 2) Browser replaces complete Flash Applet and shows message like player crashed so that also you need to restart the Browser 3) Player freezes and no more clicks possible? D) Is there a screenshot from such a crash somewhere? E) Is there something is the servers log when the crash happens? Did the crash happen always at a certain user action (B) might be the most important question. But it is also not clear to me yet what kind of crash we are talking about. Flash Player crash can have so many different meanings. Thanks! Sebastian Am 14.03.2013 02:43 schrieb Maxim Solodovnik solomax...@gmail.com: It seems like we have proof swf11+h264 not causing performance degradation @Sebastian what do you think, is it time for RC2? On Wed, Mar 13, 2013 at 7:56 PM, Alexei Fedotov alexei.fedo...@gmail.comwrote: Great job, Irina! Thanks for digging this out. -- With best regards / с наилучшими пожеланиями, Alexei Fedotov / Алексей Федотов, http://dataved.ru/ +7 916 562 8095 On Wed, Mar 13, 2013 at 4:28 PM, Irina Arkhipets irina.arkhip...@gmail.com wrote: Hi All, I investigated flash plugin crashes today and would like to share some results. First of all, crash happens both with SWF10 and SWF 11. It seems like the problem is reproducible with H263, H264 1.2 and H264 5.1. Actually, I haven't noticed significant CPU usage degradation for H264 in compare with H263 - it always was about 50-70% when I enter OpenMeetings room. The same is true about the memory usage. I was able to get the crash report in Mozilla, however for now don't see any info which can help me there. Tested environment was: Windows XP + Mozilla FireFox 19.0 + flash player 11.5.502.146. Tested server was demo.dataved.ru, there were two users with videos (1 with big window and the other with normal) in the room + me with normal window. For now anybody can try different cases on http://demo.dataved.ru/openmeetings and get own expression. Below are the links which can be tested and my today results for each of them: - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf10.swfjunk=666- SWF10 + H263. Crashed 3 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h263.swfjunk=666- SWF11 + H263. Crashed 0 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_1_2.swfjunk=666- SWF11 + H264 1.2. Crashed 2 times from 10. - http://demo.dataved.ru/openmeetings/?swf=main.as3.swf11.h264_5_1.swfjunk=666- SWF11 + H264 5/1/ Crashed 1 times from 10. Best regards, Irina. On Wed, Mar 13, 2013 at 5:02 PM, Maxim Solodovnik solomax...@gmail.com wrote: that is definitely something not readable :( I can see the word static in the generated code :) this is global variables :))) On Wed, Mar 13, 2013 at 4:57 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: just for consideration, attached is the current AS3 that OpenLaszlo generates. https://cwiki.apache.org/confluence/download/attachments/27836308/lzswf9.tar.gz Sebastian 2013/3/13 Maxim Solodovnik solomax...@gmail.com I thought maybe it will be simple to convert swf10 folder of our sources + I would like someone else will handle it : On Wed, Mar 13, 2013 at 4:38 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry but that is useless. The AS3 code that OpenLaszlo compiles is not human readable. It is generated into some tmp directory. You can search on your disc for a file called LZC_COMPILER_OPTIONS That should be into the same directory as the actual AS3 files. The thing is that there are of course some high level similarities. However there are also high level similarities between AS3 and HTML5. For example an ActionScript3 Sprite is nearly the same as an HTML div or maybe span but you can really translate those things just and expect the functionality is the same. The entire notation in MXML is different, same about the script blogs. A script blog in openlaszlo looks like: var zxac = 3; in AS it would look like: var zxac:int = 3; No matter how you do it, you will have to go through every file, debug every line of code, understand the logic, convert it to MXML, test it. And some constructs of course won't translate that easy. For example in Flex you can't create that simple a global pointer. Like: canvas.myattribute or parent.myattribute This simply won't compile. AS3 won't let you
Re: SWF11 components performance/stability
h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559 and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone and switch to h264 According to the Google switching to h264 can cause higher CPU usage. Can this be fixed by changing H264Level? (currently H264Level.LEVEL_1_2 is used) Maybe you can us understand were performance/stability degradation caused by code changes? Or maybe it is because of Flex4.6.0? Can stability be improved by switching to Flex4.9.1? Thanks in advance for your help [1] http://markmail.org/message/2el63qarsjmbvfyy http://markmail.org/message/2el63qarsjmbvfyy-- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com
Re: SWF11 components performance/stability
@Sebastian, maybe you know, is there any possibility to automatically/semi-automatically convert our swf10 sources to apache flex xml? Since Openlaszlo seems to be not actively developed and flex is now part of Apache, maybe we can benefit from such conversion? On Tue, Mar 12, 2013 at 4:28 PM, Maxim Solodovnik solomax...@gmail.comwrote: I have forwarded all missing messages On Tue, Mar 12, 2013 at 4:25 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: you wrote me without CC'ing Irina or the list :) I missed the 'reply all' button somehow. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com this one https://issues.apache.org/jira/browse/OPENMEETINGS-559 @Irina can you describe your problem more detailed? On Tue, Mar 12, 2013 at 3:46 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: We should simply do some tests to measure the actual video with h264 and the default Sorensen Codec. Then we have some numbers. What are the exact reports for Flash Player crashing ? Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com I can add switching H264 off, but I'm not sure if this is the problem. Most claims are related to the crashing flash in browser. Do you think this would help? On Tue, Mar 12, 2013 at 2:56 PM, seba.wag...@gmail.com seba.wag...@gmail.com wrote: Sorry I did not meant default h323 I meant the default Sorenson Spark http://en.wikipedia.org/wiki/Sorenson_Spark(Sorenson H.263) Sebastian 2013/3/12 seba.wag...@gmail.com seba.wag...@gmail.com h264 simply needs more computing time. AEC also is enabled (which needs some percentages more CPU power then the previous implementation). Switching to later Flex SDK is surely desirable but basically the performance is a Flash Player/VirtualMachine issue not a Flash Compiler issue. So as long as there is no Flash Player 12 that requires a later Flex SDK it is unlikely that you will gain performance advantages. I think there is a hardware acceleration for playback of h264 video, but I have really no idea about producing h264 in the browser in that context. From my point of view the main reason for using h264 is the SIP integration. For users that don't use SIP integration using h264 will simply just mean an annoying CPU usage. The case is different for AEC from my point of view. AEC brings enhancement for everybody immediately. And the more of CPU that it uses is only 3-4 % from my point of view. So my preference would be to make the video codec use the standard h323. To be honest: Who has ever complained in the last couple of years about video quality? The discussion is almost always about audio delay, audio quality, micro echo, bad microphones et cetera but actually never about video quality. And from what I have seen h264 also brings no improvements in terms of bandwidth usage. h264 could be available via configuration (or to enable SIP integration), but h323 should be used by default. Sebastian 2013/3/12 Maxim Solodovnik solomax...@gmail.com Hello Raju, Sebastian, according to https://issues.apache.org/jira/browse/OPENMEETINGS-559and various user reports (including mail from Sebastian[1]) Main code changes were: adding EnchancedMicrophone and switch to h264 According to the Google switching to h264 can cause higher CPU usage. Can this be fixed by changing H264Level? (currently H264Level.LEVEL_1_2 is used) Maybe you can us understand were performance/stability degradation caused by code changes? Or maybe it is because of Flex4.6.0? Can stability be improved by switching to Flex4.9.1? Thanks in advance for your help [1] http://markmail.org/message/2el63qarsjmbvfyy http://markmail.org/message/2el63qarsjmbvfyy-- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- Sebastian Wagner https://twitter.com/#!/dead_lock http://www.webbase-design.de http://www.wagner-sebastian.com seba.wag...@gmail.com -- WBR Maxim aka solomax -- WBR Maxim aka solomax