=== mru_ is now known as mru
ibirisronynandy: is the codec you are packaging speex?08:18
ibirisronynandy: I have created this blueprint for your release in 1109 https://blueprints.launchpad.net/linaro-multimedia-project/+spec/engr-multimedia-speex-armport08:23
ibirisronynandy: it is still in the backlog, so please do 2 things to get it targeted for 1109: 1) write the work items, 2) update the milestone yourself to be 2011.09 :-)08:24
ibirisronynandy: it would also be good to update the description summary for that blueprint I just created as a 'vanilla' almost empty one08:27
* ronynandy is back after lunch08:40
ronynandyibiris:thx,I will start updating it right away08:40
benjiG1ibiris: why openmax reference BP is bouncing between drafting and approved ?08:42
ibirisbenjiG1: just tests09:02
ibirisI was trying to see whether a combination of setting the driver for the project/project group and modifying some fields updates the status of a blueprint09:03
ibirisshould be approved now :-)09:03
benjiG1status is "Not started"  now09:08
=== benjiG1 is now known as benjiG
ibirisbenjiG: should be started now09:13
ibirisI just updated it09:13
ibirisbenjiG: you can update also - actually you should - the implementation status, otherwise I end up taking your Launchpad karma :-D09:15
ibirisfor something I have not done myself (ie start the implementation)09:16
benjiGI don't take care of my karma :-)09:17
benjiGI need to talk with krtaylor about this BP, maybe we should postpone it ...09:18
ibirissure, or put in the backlog if you like for further investigation09:19
ibirisbut since this is 1110 material check with kurt09:19
ibiriskrtaylor: somehow my linaro-multimedia subscription is screwed up :-)09:19
ibiriskrtaylor: since yesterday thunderbird is coughing and spitting error messages related to linaro-multimedia autosync...09:20
benjiGibiris: do you know if graphic WG could be interested by a guy with good skills in android graphic layers (surfaceflinger, gralloc, copyblit, etc.. ) ?12:02
ibirisbenjiG: I can check - what is it for?12:02
ibirisah sorry I read your message in the wrong way12:03
ibirisI can check :-D12:03
* krtaylor reads scrollback13:25
krtayloribiris, I am trying to glean what BP you and benjiG were discussing - backlog or investigation?13:26
ibiriskrtaylor: sorry *should*  be going back to the backlog13:28
krtaylorok, but which BP?13:30
ibiriskrtaylor: https://blueprints.launchpad.net/linaro-multimedia-project/+spec/engr-multimedia-openmax-reference13:45
krtaylorah, ok, yes agreed, should be in backlog, maybe long term13:46
ibiriskrtaylor: moved it already13:49
ibiriskrtaylor: here is my shortlist on the ideas proposed at the brainstorming yesterday13:50
ibiris -libpng optimization13:50
ibiris - better video rendering integration in UI13:50
ibiris - Audio DTS decoding - could be tricky, involves legal aspects which13:50
ibirisneed to be carefully looked at13:50
ibiris - Compressed data sound support (as in13:50
ibiris - Realvideo on ARM (popular on countries like China) - needs13:50
ibirisoptimization for eg 720p playback (VGA is ok)13:50
ibiris - Other possibilities:13:50
ibiris   + armv6 optimisations for vp8 in libav13:50
ibiris   + 10-bit h264 optimisation needed in libav (low prio) ?13:50
ibiriskrtaylor: the last 2 items were put in the agenda a week back  by mru13:50
ibirisalso - I was thinking is there anything that can be done to get flash to play on ubuntu at least?13:51
ibiris(personal itch - my daughter saw my freescale board now connected to the TV at home and the first thing she asked: "does it work with youtube?")13:52
ibirisSince we are dealing with open source, I have heard of gnash - but not sure if that would work without extra strings attached.13:53
krtayloribiris, I don't really have a problem with flash/youtube with Ubuntu (or Android) - don't really understand the question13:54
ibiris(maybe my problem then): krtaylor does it work on your eg pandaboard?13:55
krtaylorI have a few more things for that list, but its a good start from the conversation in meeting13:55
ronynandyibiris:flash doesent work on a freescale based netbook i have running maverick13:57
ronynandyflash is problem on ubuntu on arm13:57
av500arm flash works "fine" under android13:58
ibirisronynandy: I installed on my freescale imx53 quickstart board an ubuntu LEB and flash playback did not work indeed ... But that was done only yesterday so I did not have time yet to search for bugs etc13:59
ronynandyav500: I guess so as I never head of this problme for galaxy tab13:59
av500ronynandy: this is of course because there is an android version from adobe but not a generic linux one14:00
ibirisfor the record I am testing the Linaro leb here :-) If I wanted finished product testing of flash playback I can use my other (ubuntu/android) devices14:01
ronynandythis problem is definitely worth looking into..14:01
ronynandyibiris: cool14:01
ronynandyi am sure you android tablet does the flash thing fine14:02
krtaylorok, yes, my experience is with beagleC4, now that I think about it might be with extras for the board, been a while since I played with it but that was one of my test drivers to play audio(video) from youtube14:05
ronynandyjust got some expert opinion here: fash seems to use Open Gles libraries and also video librabaies like h.264 so if a properly optimided/HW accelerator is not available on the software/HW stack it wont work14:09
ronynandyit is more of an integration issue for the product rather than an issue of unoptimised components14:10
ronynandyh.264 is a popular video decoder14:11
av500ronynandy: yes14:12
av500flash is mostly an optimization issue for flash14:13
ronynandyfolks here have ported it on chrome and android..they told me that they ported the source provided by adobe directly and it worked fine14:14
ronynandythe source is neon/arm optimised...14:14
ronynandythe drivers for the HW acclerator had to be made open max complaint14:14
* ronynandy leaving for the day14:16
krtayloribiris, and all, I am not feeling well today, but still trying to get some work done - just fyi14:30
krtaylorplease email if I don't respond here and you need to communicate something14:31
krtaylorre: requirements discussion in yesterday's meeting - there was no interest from anyone in creating a standard interface for sw codecs across android and linux15:11
av500its omx, no?15:11
av500THE industry standard15:11
* av500 hides15:11
benjiGav500: shame on you !15:12
av500krtaylor: are the notes about this?15:15
krtaylorav500, but it could be the standard, or something else, I mean, is GST the defacto standard?15:15
av500krtaylor: that is asking the wrong question15:16
benjiGrequirement discussion: do we have interest on 3D video stream and secure streaming (like a netflix)15:16
av500a codec interface should be a very simple thing15:16
krtaylorav500, I have no doubt  ;-)15:16
av500gst and omx do stuff on top of that15:16
av500like buffer management15:16
av500and plugging in and out ports15:16
av500and caps15:16
av500but all that can be on top of a sane codec API15:16
av500benjiG: netflix is hard15:17
av500krtaylor: so people write 2k lines of omx boilerplate on top of 20lines codec API15:17
av500so *every* codec15:17
benjiGav500: sure, but since netflix wants to go out of america the interest is growing here :-)15:18
av500benjiG: yes15:19
av500but atm their idea of secure means really "secure"15:19
av500full chain of trust15:19
av500secure boot15:19
krtaylorav500, right, so why are we not seeing a strong public movement to fix it? is it because it was done once and it is mostly a cut/paste to support anything new?15:20
av500I see it being done *once* a lot :)15:20
av500krtaylor: in fact, it could be done in OMX by adding one more, uhm, layer15:21
av500underneath IL15:22
av500or was it DL?15:22
av500or AL?15:22
krtaylorok, well I know khronos/openmax is working on it, I guess we'll see what is proposed15:22
krtaylorI assume you are being sarcastic here ;-)15:23
av500nah, i really forgot15:23
av500ok, between IL and DL15:23
av500IL is the 2k line crap everybody struggles with15:24
av500DL are primitives that only the codec vendor cares about15:24
av500AL is the gst replacement that will never be used15:24
av500so add CL which defines: create/control/process/destroy15:24
av500no threads, no callbacks, no verion numbers in api calls15:25
av500vendors can write to that easily and testbench it15:25
av500it can run local or via ipc15:25
av500etc. etc.15:25
krtaylorlol, interesting summary15:26
benjiGwhat a synchonous API !!! it is a blasphemy :-)15:26
benjiGI think I have already read something like that somewhere :-)15:27
av500well, create/control/process/destroy is like libav or TI CE15:27

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!