[FFmpeg-devel-irc] IRC log for 2010-06-13

irc at mansr.com irc at mansr.com
Mon Jun 14 02:00:09 CEST 2010


[00:39:27] * Terminating due to: TERM
[00:39:42] * /join #ffmpeg-devel ...
[00:39:44] *** TOPIC: Welcome to the FFmpeg development channel. | Discussions about the development of FFmpeg itself are ontopic here. | Questions about using FFmpeg or developing with the libav* libraries should be asked in #ffmpeg. | FFmpeg 0.5.2 has been released! | This channel is now publicly logged.
[00:39:44] *** TOPICINFO: DonDiego!~diego at ip-109-90-5-29.unitymediagroup.de, 1274741280
[01:43:10] <xxthink> Are there some open source bit stream analysis tools?
[02:08:36] <kierank> xxthink: h264parse
[06:24:36] <_av500_> yawn
[06:39:14] <barque> good morning/afternoon/night ladies and gentlemen
[06:39:40] <barque> I am here to speak to Mr. or Ms. kshishkov
[06:40:41] <_av500_> Mr.
[06:40:47] <barque> ahh Mr. then :D
[06:40:54] <Dark_Shikari> I wonder if he's awake
[06:40:57] <Dark_Shikari> damn ruskies
[06:41:01] <_av500_> but i guess he is travelling back from linuxtag
[06:41:05] <barque> lol it should be morning there
[06:41:06] <Dark_Shikari> ahhh
[06:41:13] <Dark_Shikari> so that's what he's up to
[06:41:24] <barque> so yeah I added alpha support to TGA?
[06:41:25] <barque> :D
[06:41:38] <barque> just umm... asking for a commit *wink*wink*
[06:41:46] <barque> =^.^=
[06:41:52] <barque> if he comes back, tell him I like Russian vodka
[06:41:59] <_av500_> it has been ack'ed?
[06:41:59] <barque> Dobre Noche Commrade Kshishkov
[06:42:07] <barque> well Dark_Shikari likes it :D
[06:42:13] <_av500_> Dark_Shikari: ukraine
[06:42:13] <Dark_Shikari> it's like 5 bloody lines
[06:42:17] <barque> 3
[06:42:20] <elenril> (actually he's not russian =p)
[06:42:20] <Dark_Shikari> ukraine, russia, same thing
[06:42:30] * elenril kills Dark_Shikari 
[06:42:54] <_av500_> Dark_Shikari: but u know Canada is not an US state?
[06:43:01] <barque> LOL I'm sure he knows
[06:43:15] * barque hails from Canada
[06:43:35] <Dark_Shikari> _av500_: of course not
[06:43:37] <Dark_Shikari> their milk comes in BAGS.
[06:43:42] <barque> :D
[06:43:44] <Dark_Shikari> obviously, they're not part of the US
[06:44:34] <Dark_Shikari> also, they don't know what bacon is
[06:44:44] <bcoudurier> send the patch to the ffmpeg-devel
[06:44:57] <Dark_Shikari> yeah, I suggested that.
[06:45:19] <bcoudurier> so all people can review it/test it
[06:45:35] <barque> okay sure thing
[06:45:37] <Dark_Shikari> it's http://pastebin.com/hewpqRuA
[06:45:39] * elenril wonders why does everybody want to edit history so much
[06:45:42] <bcoudurier> actually I'd like to test it, I use targa sometimes
[06:45:42] <Dark_Shikari> add alpha support to tgaenc
[06:45:46] <barque> oh yeah and that's the link
[06:46:23] * barque signs up for the mailing list
[06:47:02] <barque> waiting for approval :]
[06:47:44] <barque> of ... my email address :D
[07:01:10] <barque> ok I'll probably show up later or something
[07:01:21] <barque> but umm great stuff
[07:01:30] <barque> have a good night/morning/noon/evening everyone
[07:27:40] <hyc> and what's wrong with just forcing a new commit with no changes, just "previous comment was bogus ..."
[07:28:00] <hyc> i.e., just enter a new commit message
[07:28:24] <hyc> we never edit commit messages in cvs
[07:35:09] <elenril> also it's not like anybody except DonDiego ever edits commit messages
[07:36:02] <ohsix> is this the git thing? can't people just amend in their local tree & read it before submitting
[07:36:47] <elenril> commit messages are write-only, reading is too much effort
[07:36:56] <hyc> lol
[07:37:18] <ohsix> eh, when you put stuff upstream and out of your tree its kind of a big deal :]
[07:37:53] <hyc> still, I don't see why you can't just make a no-op commit with a new message
[07:38:30] <elenril> hyc: write a mail to the ml
[07:38:48] <elenril> git opponents don't come here
[07:39:07] <Yuvie> it's ugly and there might be a dozen commits since the bad message
[07:39:13] * elenril wonders if it has some deeper meaning
[07:39:15] <ohsix> opponents? wheres the thread
[07:39:28] <hyc> ugly - big deal
[07:39:36] <Yuvie> the notes are less ugly
[07:39:43] <ohsix> git isn't that big of a deal; its really just central vs. distributed
[07:39:44] <Yuvie> but I dunno when they're displayed
[07:39:47] <hyc> as long as eventually the history is recorded...
[07:40:01] <elenril> git is a big deal
[07:40:06] <elenril> git >>>>> svn
[07:40:14] <elenril> </fanboyism>
[07:40:16] <ohsix> in git you'd probably rv and propose to recommit if it actually gets upstream somewhere with a bad message
[07:40:52] <elenril> you can always just revert it and commit again if it was _that_ bad
[07:40:54] <ohsix> stuff could should picked between branches anyways; could fix messages during that transition too
[07:41:30] <hyc> picked instead of pulled/merged?
[07:41:52] <ohsix> i'm just trying to say theres a bunch of ways to QA commit messages; git if anything makes it way easier to do in your tree if you make a mistake
[07:42:06] <Yuvie> but really, commit message editing isn't a big deal imo, most of what I've seen edited either typos or simple commits that someone wrote something like "fix this"
[07:42:43] <hyc> eh, fixing in your own tree is pretty pointless - if you wrote a bad commit msg, who is going to see it to tell you to fix it?
[07:42:53] <hyc> no one knows until you've pushed it...
[07:43:12] <Yuvie> but you have to fix it in your own tree before pushing it...
[07:43:25] <ohsix> if you let shit in your own tree ...'
[07:43:41] <hyc> that is the obvious problem people are worried about
[07:43:51] <ohsix> and changes tend to go between branches in sets, it makes a lot of sense you'd review the large portion before it touches anything
[07:44:39] <hyc> still... if you have write access to the public repo, no one reviews your commit msg until you push.
[07:44:42] <Yuvie> there's a fair amount of ffmpeg commits that don't get reviewed beforehand
[07:44:45] <ohsix> the tree that nobody wanted typos in could be managed by itself; during rfcs or could just be flat rejected by some written standards
[07:44:58] <Yuvie> maybe that would change if everyone did 'git push' instead of 'svn commit'
[07:45:03] <ohsix> that sounds like a problem that would exist regardless heh
[07:45:44] <hyc> I guess you could do as the kernel does - no pushes
[07:45:50] <ohsix> in the git world stuff exists in personal trees for some time; and people make them public to pull from, stuff only hits the main tree after that; ideas are cheap to test and discard
[07:45:51] <hyc> just pull requests sent to linus
[07:46:22] <hyc> so you designate one person in charge of maintaining the repo, who handles all pull requests
[07:46:43] <hyc> and then it's their problem to review all commit msgs
[07:46:51] <ohsix> for the main tree, ya
[07:47:14] <ohsix> but even upstream can have branches you can push to, and ones you cant
[07:47:45] <ohsix> might put tinderbox on something that gets open commits; let people do bad things there, let it hit the main tree by request or inspection
[07:47:53] <hyc> that would work too
[07:48:11] <ohsix> doesn't make a lot of sense for badness to make it upstream though; public personal trees are easy
[07:48:55] <hyc> public trees yes, but it still means someone else has to examine your tree before your code goes anywhere upstream
[07:49:04] <Yuvie> is tinderbox anything like gerrit?
[07:49:06] <ohsix> of course
[07:49:07] <hyc> so any process along these lines doubles the work
[07:49:27] <ohsix> Yuvie: fwiw i wasn't suggesting "tinderbox" persay; for ffmpeg that would be fate
[07:49:58] <Yuvie> (libvpx's commit history is littered with useless merge commits, for which I blame gerrit)
[07:50:00] <ohsix> hyc: i dunno; seems the same to me, unless other persons are interested in actively tracking personal trees
[07:50:25] <ohsix> but personal trees might not even exist until a review request, not unlike a patch might recieve
[07:52:12] <ohsix> the concern about commits to the One True Tree(tm) are really different with a dvcs and insanely cheap branches
[07:52:28] <hyc> so it would be the equivalnt of just emailing git format-patch to the list
[07:52:43] <ohsix> speaking in terms of the restraints of the other is kind of useless; you're arguing for what you have now, not what the new one might offer
[07:53:11] <ohsix> sure, or pull requests if someone wanted to moderate/handle those, the patch might just be for an rfc, or it might be what gets comitted; you can do what you want there though
[08:06:13] <_av500_> most patches are reviewed anyway, so make the commit msg part of the review...
[08:07:26] <pross-au> and add ispell pre-commit hook
[08:07:53] <_av500_> yup
[08:08:25] <_av500_> SMS to DonDiego pre-commit hook
[08:08:30] <hyc> who cares about spelling errors in commit msg? as long as there are no typos in the actual commtited code...
[08:08:45] <hyc> and people can read the msg
[08:09:10] <ohsix> well it just kind of shows a lack of care in preparing it; it might not matter for simple commits but it will matter more when features come whole hog from other branches/trees
[08:09:36] <ohsix> don prepares the changelog doesn't he? he might be extra sensitive huhuhu
[08:10:21] <pross-au> +1 for that av500
[08:11:05] <pross-au> But yeah, engineers love a good bikeshed
[08:15:11] <_av500_> at work, i use "icc" often as commit message
[08:15:45] <_av500_> aka "increase commit count"
[08:16:17] <hyc> what, are you paid by the number of commits?
[08:18:27] <_av500_> hyc: yeah :)
[08:18:34] <_av500_> well, no :(
[08:18:55] <hyc> ;)
[08:20:44] <Gottaname|Mobili> anyone used ffserver before?
[08:20:52] * Gottaname|Mobili can't believe if himself or ffserver that's fried
[08:21:13] <hyc> ffserver is a steaming pile of ... almost works
[08:21:25] <hyc> working great for me now with the patches I've posted
[08:22:39] <Gottaname|Mobili> hyc: well, is it able to stream from file?
[08:22:43] <Gottaname|Mobili> it worked... almost.
[08:22:50] <hyc> depends on the file
[08:22:54] <Gottaname|Mobili> flv
[08:23:02] <hyc> no, not without my patches
[08:23:07] <Gottaname|Mobili> setup feed, then use ffmpeg to pump the feed
[08:23:09] <Gottaname|Mobili> and segfault
[08:23:09] <Gottaname|Mobili> :D
[08:23:32] <hyc> what exactly are you doing? what codecs?
[08:23:32] <Gottaname|Mobili> hyc: patches already made to svn?
[08:23:52] <Gottaname|Mobili> just trying to get an flv file to stream
[08:23:57] <Gottaname|Mobili> x264, mp3
[08:24:10] <hyc> no, they're being held up... people seem to dislike the color/font of my patches.
[08:24:26] <Gottaname|Mobili> fook, while its killing those who need it working
[08:24:37] <Gottaname|Mobili> anyway, correct me if I'm wrong
[08:24:42] <hyc> (j/k. they're held up for mostly irrelevant reasons, from what I see...)
[08:25:01] <Gottaname|Mobili> if you intend to say stream some mp4 file, you still need to setup a feed and have ffmpeg convert to that feed file right?
[08:25:15] <hyc> you can stream from a file directly
[08:25:28] <hyc> you only need to use a feed file if you want to push something live into it
[08:25:28] <Gottaname|Mobili> yeah. but its ghey, file tries to download itself whole
[08:25:59] <Gottaname|Mobili> any secret ffserver config I don't know about to force the file to stream and not download in whole?
[08:26:06] <hyc> are you trying to stream over http or rtsp?
[08:26:18] <Gottaname|Mobili> tried both, didn't work
[08:26:32] <hyc> I don't understand your question. stream vs download...
[08:26:41] <Gottaname|Mobili> maybe I'll pastebin
[08:26:43] <Gottaname|Mobili> 1 sec
[08:28:10] <Gottaname|Mobili> http://ffmpeg.pastebin.com/FMkgyq4f
[08:28:28] <Gottaname|Mobili> technically the above should stream correctly?
[08:28:35] <hyc> right, that won't work at all without my patches
[08:28:46] <Gottaname|Mobili> can I have them? pretty please
[08:28:46] <Gottaname|Mobili> :(
[08:29:34] <hyc> trying to find the relevant bit, you can start here https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-May/thread.html
[08:29:44] <hyc> scan down for ffserver on that page
[08:30:13] <Gottaname|Mobili> ah the global header bug right?
[08:30:21] <hyc> that's the start of it
[08:30:27] <hyc> some of those patches have been committed
[08:30:44] <hyc> trying to find the part that's directly relevant here
[08:30:49] <Gottaname|Mobili> since when?
[08:30:59] <hyc> a couple days ago
[08:31:04] <Gottaname|Mobili> holy
[08:32:50] <Gottaname|Mobili> its been committed to svn?
[08:33:44] <_av500_> sadly yes
[08:33:50] <_av500_> git ftw!
[08:35:35] <hyc> this one https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-May/088834.html
[08:35:43] <hyc> is still not fixed in svn yet
[08:35:52] <hyc> and without it flv will not work
[08:37:24] <hyc> i think all in all I ran into 15 or 16 separate bugs in ffserver code
[08:37:38] <hyc> most of the fixes are in svn now
[08:37:39] <Gottaname|Mobili> of which how many are still remaining?
[08:37:49] <Gottaname|Mobili> hmm, I should do another svn update
[08:37:52] <hyc> but this bit and one or two others still remain
[08:38:27] <Gottaname|Mobili> how about mp4 RTSP?
[08:38:52] <hyc> I haven't checked. I would assume it's ok.
[08:39:28] <Gottaname|Mobili> hmmm
[08:39:36] <Gottaname|Mobili> I'll do another check and try again
[08:41:50] <hyc> speed of getting ffserver fixes in is pretty slow. Baptiste Coudurier is the maintainer and he's usually busy with other code.
[08:42:35] <Gottaname|Mobili> acutally, is the 5.1 stable release working?
[08:42:49] <hyc> what 5.1 ?
[08:43:05] <Gottaname|Mobili> the current stable
[08:43:20] <hyc> not likely
[08:43:21] <Gottaname|Mobili> oh wait
[08:43:30] <hyc> that's from 2009
[08:43:40] <Gottaname|Mobili> oh
[08:43:48] <Gottaname|Mobili> okay, crap
[08:44:01] <Gottaname|Mobili> I'll guess I'll have to grab the update from svn and try again
[08:44:21] <hyc> it's 0.5.1, not 5.1 ...
[08:45:21] <Gottaname|Mobili> yeah, my brain is abit fried atm
[08:45:26] <Gottaname|Mobili> whoops...
[08:59:51] <KotH> morgä
[09:05:27] <CIA-92> ffmpeg: stefano * r23592 /trunk/ (libavutil/avutil.h doc/APIchanges): Add APIchanges entry and bump minor after av_compare_mod() addition.
[09:07:37] <CIA-92> ffmpeg: stefano * r23593 /trunk/libavutil/mathematics.h: Fix av_compare_mod() doxy.
[09:25:38] <Gottaname|Mobili> hyc: I've updated my ffmpeg version, but ffserver still won't handle mp4 RTSP or stream from file
[09:27:57] <CIA-92> ffmpeg: stefano * r23594 /trunk/libavfilter/parseutils.c: Add missing check to av_get_token().
[09:31:57] <hyc> dunno what to tell you
[09:32:52] <CIA-92> ffmpeg: stefano * r23595 /trunk/libavfilter/parseutils.h:
[09:32:52] <CIA-92> ffmpeg: Resort @param in av_get_token() doxy to reflect the same order they
[09:32:52] <CIA-92> ffmpeg: have in the function.
[09:32:52] <CIA-92> ffmpeg: stefano * r23596 /trunk/libavfilter/parseutils.h: Document what av_get_token() returns in case of allocation failure.
[09:33:55] <hyc> you updated and patched?
[09:34:03] <Gottaname|Mobili> yeah, latest svn
[09:34:35] <hyc> mp4 is kind of problematic
[09:35:20] <hyc> the need for a moov atom means you can't use it as a live format
[09:35:34] <Gottaname|Mobili> any format which you think is a safe bet?
[09:35:36] <Gottaname|Mobili> mpg?
[09:35:52] <hyc> flv works with my patches
[09:36:56] <Gottaname|Mobili> yeah, it now works partially
[09:37:09] <Gottaname|Mobili> let me verify what's happening..
[09:37:57] <Gottaname|Mobili> weird...
[09:38:45] <Gottaname|Mobili> http://ffmpeg.pastebin.com/rUW9SerQ
[09:39:36] <Gottaname|Mobili> file downloads but is unabled to be played while being dled
[09:40:36] <hyc> that was with an flv file?
[09:40:55] <Gottaname|Mobili> yep
[09:41:25] <Gottaname|Mobili> using streaming from file
[09:41:27] <hyc> then there's something wrong with the file
[09:41:32] <Gottaname|Mobili> hmmm
[09:41:36] <hyc> flv files always contain global headers
[09:41:47] <Gottaname|Mobili> okay, I'll go hunt for another
[09:41:53] <Gottaname|Mobili> do I need to set the format to rtp?
[09:42:16] <hyc> only if you're going to stream by rtsp
[09:42:24] <hyc> if you want to use http, then no
[09:44:32] <hyc> probably that flv file is not using compatible codecs
[09:44:44] <hyc> H.264 / AAC will work.
[09:47:04] <Gottaname|Mobili> hyc: it works in a sense it downloads the whole file and plays
[09:48:01] <hyc> what downloads? what are you using as a client?
[09:48:40] <Gottaname|Mobili> I have VLC, wmp, wmp classic
[09:49:02] <hyc> it streams in realtime for me, using rtsp client on my G1 phone
[09:49:10] <hyc> also streams using ffplay
[09:49:23] <Gottaname|Mobili> hyc: I must be doing something wrong
[09:49:41] <Gottaname|Mobili> let me check again, I need to setup as a feed, or stream from file or use format rtp?
[09:49:52] <hyc> maybe some other patches are missing still. there's nothing special to stream an flv file over rtp
[09:50:29] <Gottaname|Mobili> that's what I think..
[09:50:52] <Gottaname|Mobili> anyway I could have your copy off ffserver.c?
[09:51:10] <hyc> that's not the only part
[09:51:16] <hyc> you also need libavformat fixes
[09:52:25] <hyc> but those should all have been in the patch I pointed you at
[09:52:47] <Gottaname|Mobili> https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-May/088678.html
[09:52:50] <Gottaname|Mobili> this one right?
[09:52:53] <hyc> yeah
[09:53:15] <Gottaname|Mobili> https://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-May/088834.html or this one?
[09:53:32] <hyc> that last one 88834
[09:57:05] <Gottaname|Mobili> reinstalling...
[10:09:44] <enkidu> Gottaname|Mobili: dont worry, I had to reinstall whole system, because configure script is not checking for library compatibiliity ~~
[10:10:17] <Gottaname|Mobili> yeah urgh...
[10:10:35] <hyc> reinstalling what?
[10:10:44] <hyc> not just recompiling ffmpeg?
[10:10:56] <enkidu> hyc: I had to reinstall system, as I had broken libs
[10:11:02] <hyc> ugh
[10:11:14] <enkidu> strange was the fact, they worked with 0.5.2 snapshot
[10:11:14] <Gottaname|Mobili> okay done...
[10:11:16] <Gottaname|Mobili> testing...
[10:12:13] <Gottaname|Mobili> this better work or I will be seriously pissed off...
[10:12:57] <hyc> I know the feeling
[10:12:59] <Gottaname|Mobili> no good, still getting the same global header errors
[10:13:10] <hyc> what does ffprobe on your flv file say?
[10:13:20] <Gottaname|Mobili> let me check
[10:13:54] <enkidu> Gottaname|Mobili: do you have AVOptionVideo +global_header ?
[10:14:07] <hyc> don't need that when streaming a file
[10:14:07] <Gottaname|Mobili> no
[10:14:11] <hyc> only need that for feeds
[10:14:18] <enkidu> ah, ok
[10:14:42] <Gottaname|Mobili> http://ffmpeg.pastebin.com/LFBxxTWm
[10:14:53] <Gottaname|Mobili> hyc, did you use file or feed streaming?
[10:14:57] <hyc> both
[10:15:10] <hyc> your file uses flv vcodec and mp3 acodec
[10:15:16] <hyc> I told you only h264 / aac will work
[10:15:26] <Gottaname|Mobili> oh... fug...
[10:15:47] <enkidu> hyc: mp3 audio is NOT supported? why?
[10:16:02] <hyc> for rtp? dunno, maybe it will work
[10:16:06] <hyc> flv definitely will not
[10:16:21] <enkidu> yeah, flv video is only for http
[10:17:24] <enkidu> anyways, I am fighting dependences
[10:17:36] <Gottaname|Mobili> hyc: maybe I can use h264 and aac to stream in the conf file?
[10:17:54] <enkidu> Gottaname|Mobili: for feed you can
[10:18:06] <enkidu> for file - you better recode
[10:18:17] <Gottaname|Mobili> I'll give it a try
[10:18:18] <Gottaname|Mobili> 1 sec
[10:18:19] <hyc> yep, you need to recode if you want to use a file
[10:18:39] <hyc> for a feed you have to configure all the codec parameters
[10:18:59] <hyc> check the doc/ffserver.conf, I added a new example towards the end
[10:19:13] <hyc> and that is already in svn...
[10:20:27] <Gottaname|Mobili> okay... using ffmpeg to reencode...
[10:20:39] <enkidu> hyc: is it possible to embed in ffserver some management code?
[10:20:47] <hyc> what management
[10:21:10] <enkidu> killing clients etc
[10:21:31] <enkidu> reloading conf file
[10:21:40] <hyc> I suppose anything's possible, but that would be some work
[10:21:48] <Gottaname|Mobili> actually
[10:21:59] <hyc> controlled thru the http status page perhaps
[10:22:04] <Gottaname|Mobili> I'm trying to improve on ffserver, or build something equivalent
[10:22:17] <enkidu> I know it. Anyways, I have to rewrite my css patch for ffserver
[10:22:32] <KotH> uhm.. how do i add a comment in roundup?
[10:22:33] <hyc> there are other servers already out there, like darwin or feng
[10:22:37] <Gottaname|Mobili> okay, re-encoding with -vpre slow*
[10:22:39] <Gottaname|Mobili> yeah
[10:22:42] * KotH seems to be either blind or very stupid
[10:22:45] <Gottaname|Mobili> but they're complex
[10:22:51] <Gottaname|Mobili> ffserver is dead simple
[10:23:15] <hyc> mostly ffserver is dead. it needed a lot of work to make it usable for me :P
[10:23:29] <Gottaname|Mobili> I noticed, alot of things here are broken
[10:23:40] <enkidu> Gottaname|Mobili: which ones?
[10:23:54] <Gottaname|Mobili> either developers are too busy with new features till it leaves a giant trainwreck of support and documentation headaches
[10:24:08] <Gottaname|Mobili> the way stuff are depreciated so fast
[10:24:16] <Gottaname|Mobili> like avcodec_decode_audio and such
[10:24:25] <enkidu> yeah...
[10:24:32] <Gottaname|Mobili> dranger's tutorial and such has become irrelevant
[10:24:46] <Gottaname|Mobili> it affects the ability for people to pick up ffmpeg API usage
[10:24:54] <enkidu> I can try some work on ffserver
[10:25:03] <hyc> well, the existing apis frequently suck
[10:25:16] <hyc> they need to be overhauled anyway
[10:25:27] <Gottaname|Mobili> hyc: tell me if this is better...
[10:25:27] <Gottaname|Mobili> 1 sec
[10:25:50] <Gottaname|Mobili> http://ffmpeg.pastebin.com/LudnwL9V
[10:26:04] <hyc> that looks like it should work
[10:26:07] <Gottaname|Mobili> okay
[10:26:11] <Gottaname|Mobili> in it goes...
[10:27:02] <Gottaname|Mobili> crap
[10:27:03] <Gottaname|Mobili> no go
[10:27:10] <Gottaname|Mobili> still dling as a whole
[10:27:21] <Gottaname|Mobili> still giving global headers error...
[10:27:36] <hyc> you sure it's reading the right file?
[10:27:41] <hyc> did you restart ffserver?
[10:27:46] <Gottaname|Mobili> yeah
[10:29:29] <Gottaname|Mobili> bin the conf file...
[10:29:55] <Gottaname|Mobili> http://ffmpeg.pastebin.com/R3PgHdqz
[10:30:30] <hyc> huh?
[10:30:41] <hyc> what do you want this thing to do?
[10:31:15] <Gottaname|Mobili> play while downloading?
[10:31:57] <hyc> then your players are doing something wrong
[10:32:14] <hyc> since ffserver serves anything over http just fine
[10:32:33] <Gottaname|Mobili> having your browsers dl the file first then play is incorrect right?
[10:32:44] <hyc> nope. that's what browsers do.
[10:33:13] <hyc> only custom plugins will play progressively
[10:34:28] <Gottaname|Mobili> hyc: so I'll need a player that supports streaming right?
[10:34:33] <hyc> yeah
[10:34:36] <Gottaname|Mobili> crap
[10:34:40] <Gottaname|Mobili> any recommendations
[10:34:59] <hyc> well, you already said you have vlc
[10:35:02] <hyc> that should work fine
[10:35:17] <Gottaname|Mobili> maybe its because I'm tunneling to my local network...
[10:35:25] <Gottaname|Mobili> nvm, I'll get back upstairs
[10:35:30] <hyc> the point is, you can't let the browser open the link, you have to just give the link to the player
[10:35:31] <Gottaname|Mobili> brb in 5 minutes
[10:35:37] <Gottaname|Mobili> yeah
[10:35:46] <Gottaname|Mobili> I'll get back on my local LAN
[10:35:47] <Gottaname|Mobili> bbl
[10:37:23] <mru> this is weird... I'm useing wifi at the airport without having provided any credentials or card numbers
[10:37:34] <lu_zero> interesting
[10:37:37] <KotH> it's called security ;)
[10:37:46] <lu_zero> which a
[10:37:50] <lu_zero> ccess point?
[10:38:02] <mru> lu_zero: Hotspot-TXL
[10:38:07] <lu_zero> neat
[10:38:10] <mru> chose vodafone as provider
[10:38:23] <mru> there was a choice of credit card too
[10:38:31] <mru> surely vodafone can't identify my laptop
[10:38:52] * lu_zero is thinking about what to do for the next hours
[10:39:15] <lu_zero> outside is gloomy, the hotel lobby is confortable...
[10:39:20] <mru> sounds boring
[10:39:37] <mru> are any museums open today?
[10:39:38] * lu_zero wanted to take some proper pictures around
[10:39:53] <lu_zero> I was thinking about going to the photography museum
[10:40:09] <mru> haven't been to that one
[10:40:13] * lu_zero is quite undecided
[10:40:30] <mru> indecision is the mother of bikesheds
[10:41:32] <Gottaname> back
[10:41:35] <mru> font
[10:41:37] <mru> front
[10:42:08] <KotH> bcoudurier: is it worth to keep your [Taka]_Naruto_Shippuuden_148_[720p][E50E7705].mp4 sample ?
[10:42:37] <Gottaname> its naruto
[10:42:45] <Gottaname> its not worth keeping
[10:42:46] <Gottaname> >:3
[10:42:49] <KotH> hehe
[10:44:26] <lu_zero> pfff
[10:44:56] <Gottaname> hyc: it vorks!
[10:45:24] <KotH> lu_zero: dont tell me you're a narutard? ;->
[10:46:09] * KotH needs faster disks
[10:46:13] * KotH needs more disks too
[10:46:24] * mru watches people queue at the gate
[10:46:49] <Gottaname> my 1TB RAID5 is sufficient for me
[10:46:50] <Gottaname> :D
[10:47:04] <KotH> raid5 is too slow
[10:47:14] <KotH> and 1tb is not even barely enough
[10:47:27] <Gottaname> I am a simple man
[10:47:28] <Gottaname> :(
[10:47:38] <Gottaname> also, raid array built out of WD blacks
[10:47:39] <Gottaname> =3
[10:48:20] <KotH> lol
[10:48:26] <KotH> slow disks are slow
[10:48:34] <spaam> KotH: buy 24 ssd and use them in raid0 . then you can write 2gigabyte / sec ;)
[10:48:58] <KotH> spaam: ssd's are too small
[10:49:04] <KotH> the largest one available is 500g
[10:49:10] <mru> use many
[10:49:20] <KotH> not enough space in case
[10:49:33] <mru> enlarge case
[10:49:48] <KotH> not enough pci slots to accomdate for all the sata cards i'd need
[10:50:07] <KotH> 12 disks is max
[10:50:07] * Gottaname drops a SAN on KotH
[10:50:09] <mru> pce expander
[10:50:13] <mru> pci
[10:50:18] <enkidu> which ffpreset uses resolution of 384x288?
[10:50:29] <spaam> Gottaname: SAN is slow :)
[10:50:40] <enkidu> argh, sorry
[10:50:46] <KotH> Gottaname: i already have a NAS
[10:51:04] <mru> turn it backwards
[10:51:12] <Gottaname> NAS are for wussies
[10:51:18] <twnqx> spaam: why would SAN be slow? i like my 8gbit fiber channel
[10:51:23] <Gottaname> a real high grade enterprise SAN
[10:51:30] <Gottaname> that will set you back
[10:51:31] <Gottaname> 100k
[10:51:35] <twnqx> lol
[10:51:50] <twnqx> 4gbit FC SANs, refurbished HP, 2.5k + SATA disks
[10:52:01] <Gottaname> fuck yeah FIBRE CHANNEL
[10:52:08] <twnqx> (optionally SAS if you need higher speeds)
[10:53:09] <enkidu> i got only 1gbit :(
[10:53:16] <spaam> twnqx: not everyone has 8gbit fiber channel at home :)
[10:53:34] <enkidu> but gonna have 10gbe
[10:53:51] <twnqx> so use 1gbit iscsi, still not much slower than local discs
[10:54:03] <twnqx> or bundle to 2-4 gbit
[10:54:09] <lu_zero> 12:43 < KotH> lu_zero: dont tell me you're a narutard? ;->
[10:55:12] <lu_zero> one of my friend managed to hook me to the manga, she is translating it and her footnotes made it less naive
[10:55:50] <lu_zero> the anime IMHO is trash
[10:56:05] <KotH> ah.. it's ok if you are reading the manga
[10:56:26] <lu_zero> thehehe
[10:57:18] <lu_zero> the author isn't that bad at writing stories probably written as a book it would remain nice
[10:57:21] <KotH> btw: are there any good GUI email clients out there that can handle imap folders with >60k mails efficiently and support japanese input?
[10:57:51] <KotH> sylpheed is kind of slow (due to the use of gtk stuff)
[11:02:31] <spaam> KotH: i guess thunderbird is slow also.
[11:04:09] <KotH> slow doesnt describe it ^^'
[11:07:07] <spaam> use mutt? :)
[11:07:21] <spaam> oh you wanted gui..
[11:07:27] <KotH> yes
[11:10:49] <KotH> wtf... we've 5! uploads of the same file
[11:11:13] <spaam> Nice
[11:11:28] <spaam> backup for the backup  :)
[11:17:16] <enkidu> I think of multilayered ffserver instalation
[11:17:48] <enkidu> haproxy <-> 10 parallel ffservers <-> ffserver <-> encoders
[11:57:40] <lu_zero> claws?
[11:58:09] <lu_zero> enkidu: really, I want to know what you are up to and how it fares =)
[12:00:58] <enkidu> lu_zero: just want to avoid situation, when frontend ffserver dies and then encoders dies too
[12:01:45] <enkidu> anyways, seems, that new profiles are messing up x264 for jwplayer
[12:02:15] <lu_zero> enkidu: using http pseudo?
[12:02:33] <enkidu> yeah
[12:02:47] <enkidu> for some reason, video is not displayed
[12:02:51] <lu_zero> how many cpus?
[12:02:56] <enkidu> one
[12:03:06] <lu_zero> how many clients attached?
[12:03:37] <enkidu> doesnt matter
[12:03:39] <lu_zero> maybe the profile you are using isn't supported by flash
[12:03:49] <enkidu> maybe... which ones are?
[12:39:56] * KotH feels like he has to punish some people
[12:55:31] <_av500_> 5! uploads? thats 120, no?
[12:55:48] * KotH punishes _av500_ 
[12:57:00] * _av500_ hides in the DACH region from KotH 
[13:00:10] <twnqx> can the flac encoder keep 24bit? :X
[13:02:39] <jai> i dont think the encoder supports >16bit samples
[13:02:58] <twnqx> too bad, the original libflac does...
[13:03:35] <_av500_> patches welcome
[13:03:44] <twnqx> however, ffmpeg seems to compress far superior. at least that's the onlyidea i get if 24bit flac -8 takes 1.7GB and ffmpeg -aq 8, which converts to 16bit, need 740MB...
[13:04:49] <jai> j-b was looking for comparisons too :)
[13:05:02] <jai> twnqx: it's still faster than libflac right?
[13:05:28] <twnqx> i didn't time it, and encoding time is not in my list of constraints...
[13:05:37] <twnqx> given that the video encode runs with ~3fps :P
[13:06:41] <jai> ah
[13:06:55] <jai> well, shouldn't be that complicated to support higher bitdepths
[13:09:17] <pentanol> twnqx hm, oddy, but in my tree, it a bit outdated sample_size_table { 0, 8, 12, 0, 16, 20, 24, 0 };
[13:10:10] <twnqx> .sample_fmts = (const enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE}, for the encoder
[13:10:32] <jai> the table is in the decoder
[13:10:47] <jai> *used in
[13:11:24] <twnqx> hm, -aq is not the compression level, is it?
[13:11:58] <twnqx> ah. -compression_level, who would have thought :P
[13:16:35] <pentanol> it's  audio quality
[13:30:40] <KotH> does anyone have an idea what nrk_aac_he2 is ?
[13:30:58] <KotH> if not, the file is going to be deleted
[13:31:09] <kshishkov> sounds like AAC sample with PS :P
[13:31:14] <kierank> from nrk
[13:31:28] <kshishkov> ask Alex
[13:31:37] <kshishkov> (aka aconverse)
[13:33:12] <KotH> and he's offline, ofc
[13:34:36] <jai> broadcast stream?
[13:35:01] <kierank> iirc nrk use latm
[13:41:29] <KotH> damn.. ffmpeg developers are lazy
[13:41:55] <elenril> KotH: http://tvtropes.org/pmwiki/pmwiki.php/Main/DepartmentOfRedundancyDepartment
[13:42:00] <KotH> all the old stuff in incoming are fucking old ffmpeg bugs nobody cared to move to the right place
[13:42:13] <KotH> elenril: be carefull.. i'm not in a good mood anymore
[13:42:30] * elenril fills himself with care
[13:48:47] * KotH shots elenril anyways
[13:51:23] <elenril> http://tvtropes.org/pmwiki/pmwiki.php/Main/MoralEventHorizon
[13:55:00] <KotH> be glad i didnt use a rusty spoon
[13:56:52] <elenril> http://tvtropes.org/pmwiki/pmwiki.php/Main/ForTheEvulz
[14:07:30] * KotH is currently very tempted to do a rm -rf incoming
[14:08:52] <mru> mv incoming outgoing
[14:11:58] <kierank> KotH: does DNXHD-DOLBYE-MXF-OMNEON exist in /incoming?
[14:12:42] <KotH> no
[14:13:03] <KotH> correction, it does
[14:13:45] <KotH> and if you want that it remains available, i advise you to tell me where to move it
[14:13:56] <KotH> otherwise it'll be gone
[14:15:52] <kierank> can you make a directory for dolby E in samples please?
[14:16:07] <kierank> and stick it in there?
[14:16:08] <KotH> tell me the exact path you want to have
[14:16:54] <kierank> MXF/DolbyE/DNXHD-DOLBYE-MXF-OMNEON/
[14:17:51] <KotH> done
[14:18:46] <kierank> thanks
[14:19:02] <kierank> it's 403 forbidden
[14:19:30] <KotH> it'll be fixed in about an hour
[14:20:07] <kierank> ok
[14:21:32] <Vitor1001> Does anyone knows if one can write and test ARM neon assembler on qemu?
[14:21:44] * Vitor1001 thinks mru might know the answer
[14:27:00] <_av500_> Vitor1001: yes and no
[14:27:14] <_av500_> it fails to run complex neon stuff
[14:27:25] <Vitor1001> Then it is a pointless exercise...
[14:27:32] <_av500_> i tried to use it to prototype neon stuff and failed
[14:27:45] <_av500_> but that was 6mo ago, maybe it improved
[14:27:52] <Vitor1001> Neon stuff that would work on real hardware?
[14:27:54] <_av500_> arm sits one a boatload of qemu patches
[14:28:20] <_av500_> but cannot release them to cs or so...
[14:28:28] <Vitor1001> :p
[14:28:53] <Vitor1001> Well, anyway a sheevaplug costs only 99$
[14:29:19] <_av500_> arr, that has no neon...
[14:29:30] <Vitor1001> Ow
[14:29:31] <Vitor1001> :p
[14:29:39] <_av500_> buy a BB
[14:29:40] <kshishkov> BB!
[14:29:51] <_av500_> kshishkov: you lag!
[14:29:56] <Vitor1001> Yeah, only $149
[14:30:35] <kshishkov> _av500_: yes, my irssi runs in another country
[14:31:11] <Vitor1001> Maybe we should refer to a beagle board box
[14:31:22] <Vitor1001> BBB is an very unused acronym
[14:31:43] <_av500_> we could have BBB run BBB on a BBB?
[14:32:01] <kshishkov> _av500_: why nobody thought about taking BealeBoard sheets from TI booth and placing them next to the Beast on the first day of LinuxTag?
[14:32:28] <kshishkov> _av500_: it all depends on Ronald "BBB"
[14:32:55] <_av500_> kshishkov: TI did not pay for it...
[14:34:04] <kshishkov> _av500_: they had to listen to you, isn't that enough?
[14:34:31] <_av500_> yes. came back with bleedin ears and hearts...
[14:34:36] <_av500_> +they
[14:34:47] <janneg> and gave two beagle boards for the beast
[14:35:08] <_av500_> the BBB?
[14:35:23] <KotH> Dark_Shikari: is incoming/ark_x264_green_bars worth to be kept, if yes, please add a describing .txt
[14:35:38] <_av500_> KotH: easy: it showss green bars
[14:35:54] * KotH stabs _av500_ with a chocolate knife
[14:35:55] <janneg> _av500_: it wasn't crowded saturday but we were occupied
[14:36:07] <_av500_> bikeshed?
[14:38:22] <janneg> _av500_: the beast is in my apartment. No how/when we get it to darmstadt. If you need your beagle boards I can send them by mail
[14:39:05] <_av500_> janneg: bbl...
[14:39:57] <_av500_> err, bbbl...
[14:43:53] <kshishkov> BBB = The B using 6 BB to display BBB
[14:46:14] <janneg> kshishkov: arrived save home? How was Berlin?
[14:48:33] <kshishkov> janneg: the train arrived late by half an hour which was strange but no problems with that.
[14:48:42] <kshishkov> as for Berlin - yes, sucky place
[14:49:00] <kshishkov> I bet the capital of Prussia is significantly better
[14:49:20] * kshishkov liked museum of musical instruments though
[14:53:47] <janneg> kshishkov: the other one or just at a different time?
[14:54:39] <janneg> and we have here some museums with nice stolen objects
[14:55:25] <kshishkov> I thought those stolen objects were re-stolen by Soviet Union later
[14:55:46] <kshishkov> janneg: I meant the real capital of Prussia - Potsdam
[14:57:27] <janneg> kshishkov: not all and germany also bought some back from the russia in the nineties
[14:58:06] <kshishkov> well, if you have receipts then it's not stolen anymore :)
[14:59:25] <kshishkov> BTW, before 1940s Kharkov Museum of Art had very impressive collection but guess where it ended by 1950?
[15:01:19] <janneg> Moscow? I though you might ment Königsberg/Kaliningrad. Potsdam was just the residence of the prussian kings
[15:01:42] <kshishkov> exactly
[15:02:16] <kshishkov> BTW, now Moscow is ruled from some town next to it, so it's the same
[15:02:24] <janneg> it has some nice castles/palaces but also many ugly buildings in the city build after world war II
[15:03:05] <kshishkov> losers, USSR started building ugly buildings long before WWII
[15:22:06] * KotH wonders how some people are able to give the impression that they dont know shit with just a couple of sentences
[15:25:03] <kshishkov> couple? one sentence is usually enough
[15:25:09] <KotH> s/just a couple of/every/
[15:25:15] <KotH> ^^'
[15:25:38] <kshishkov> ah, right
[15:25:42] <KotH> i'm currently reading http://www.rebelscience.org/download/cosa002.pdf and dont know whether i should laugh or shake my head
[15:26:10] <kshishkov> laugh, it's better for your health
[15:26:20] <KotH> a total "i am right but nobody sees it!"
[15:28:14] * kshishkov remembers one guy proposing doing Haar transform in parallel. The only problem it used N_i / 2 threads on each next step
[15:28:37] <KotH> what he's basically suggesting is a dataflow system, but i doubt he got that himself... and i dont think that he know that these systems died out in the 60s
[15:30:37] <KotH> kshishkov: sounds like the guy who published a lengthy article in ieee computer that programmers should stop worrying about memory, assume that they have infinite amount of memory (ie not free any memory at all, allocate 10 times as much as they need, etc pp) and let the OS handle the rest. this way they could avoid all those nasty memory handling related bugs (ie about 99% of all bugs) and be way more productive
[15:38:41] <ln-> isn't that what happens with a lot of linux-related software?
[15:39:04] <KotH> only if it's hosted on gnome.org or freedesktop.org
[15:39:53] <kshishkov> KotH: so that guy basically reinvented Java, eh?
[15:40:09] <KotH> juup
[15:40:17] <KotH> but this "book" is great
[15:40:40] <KotH> he keeps on nagging at everyone and everything around computers
[15:41:30] <kshishkov> well, most publications here are made only for meeting requirements on number of published papers for degree or such
[15:41:40] <KotH> he really seems to think that he alone is smarter than any scientist or engineer in the industry
[15:41:57] <kshishkov> sounds pretty common
[15:42:03] <KotH> this is not a publication, it's a collection of blog posts
[15:43:44] <KotH> hmm.. i really think, this guy didn't get the hang of writing code in the first place, that's why he is rumbling about everything that is done today is wrong
[15:46:51] <iive> KotH: are you sure he didn't wrote Java?
[15:47:24] <iive> blah... too slow, that's already said.
[15:47:54] * KotH gives iive a cookie
[15:49:22] <KotH> I had my initial idea for COSA back in 1980. [...] Right away, even before I knew enough assembly language to write my first program, it was clear to me that computers were fundamentally flawed.
[15:49:27] <KotH> damn.. this guy...
[15:49:35] <KotH> ^^'
[15:50:02] <KotH> Soon afterward, I mastered assembly language and landed a job as a game programmer at a small
[15:50:04] <KotH> company. I tried to explain my idea to other programmers but they either did not care or thought
[15:50:08] <KotH> it was a dumb idea. I was stunned by their reaction. I just could not understand how seemingly
[16:02:46] <lu_zero> this wireless is slooooow
[16:07:38] <lu_zero> KotH: who's that nut?
[16:16:44] <mru> KotH: why are you reading that?
[16:17:14] <kshishkov> masochism?
[16:35:47] <KotH> lu_zero: see the link to the pdf above
[16:36:20] <KotH> mru: uhm... maybe voyeurism... wanting to see how far this guy would go to discredit himself
[16:36:43] <mru> someone you know?
[16:36:59] <KotH> nope
[16:37:18] <mru> then why care?
[16:37:21] <KotH> stumbled over a blog post of him and wanted to see whether he takes the ideas he mentioned there any further
[16:37:34] <KotH> aperantely he didnt
[17:05:30] <peloverde> Is fftrollbot broken? I'm missing some logs: http://lists.mplayerhq.hu/pipermail/ffmpeg-devel-irc/2010-June/
[17:10:36] <spaam> mru broke it
[17:11:30] <elenril> peloverde: could you explain in simple terms why is ffaacenc+stereo broken?
[17:12:03] <peloverde> My guess would be mid-side decision but there may be soem other general logic flaws
[17:12:10] <peloverde> I haven't really looked into it
[17:12:39] <elenril> ah, ok
[17:16:59] <peloverde> The first thing I need to do is completely disable mid-side and see what happens
[17:27:28] <elenril> then it'll encode two monos?
[17:28:20] <aszpain> I ve got problems when tryn to stream: is it a known bug? : Error parsing rc_eq ""; undefined constant or missing (
[17:30:36] * elenril should learn something about audio coding
[17:33:48] <enkidu> aszpain: remove this fgrom your config file
[17:36:48] <enkidu> devs, there is task to do
[17:38:00] <kshishkov> and not enough trolling to do it
[17:38:45] <enkidu> ive found many streams, that have 0 (zero) data rate to client
[17:39:00] <elenril> kshishkov: what, it was you who wrote aacenc?
[17:39:48] <enkidu> i even tried to get rid of them by writing small patch (checking rate, if 0 - return 1 and exit streaming loop)
[17:40:04] <enkidu> but it doesnt work for all occurences
[17:40:11] <kshishkov> elenril: yes, that was totally my fault
[17:40:31] <aszpain> enkidu what from my config file?
[17:40:33] * elenril blames kshishkov 
[17:41:29] <enkidu> aszpain: rc_eq
[17:41:37] <enkidu> it is not needed for streaming
[17:41:49] <aszpain> rc? u mean r?
[17:41:51] <aszpain> -r?
[17:41:54] <kshishkov> elenril: it's OK to blame for it indeed
[17:42:12] <enkidu> aszpain: from your ffserver.conf (or what you have named it)
[17:42:30] <elenril> kshishkov: why aren't you improving it then? =p
[17:42:48] <enkidu> AVOptionVideo rc_eq
[17:43:08] <aszpain> hmm let me check
[17:43:12] <kshishkov> elenril: got other stuff to do and here's another person doing it more skilled than me
[17:43:34] <kshishkov> elenril: http://tvtropes.org/pmwiki/pmwiki.php/Main/MichaelNiedermayerWannabe
[17:44:36] * elenril wasted enough time there today
[17:46:17] <aszpain> enkidu I dont  have any rc_eq in my ffserver.conf file... what is rc_eq? When I start the server the first time I get a segmentation fault... the next ones i Get this error
[17:46:50] <enkidu> aszpain: server shouldnt segfault
[17:46:56] * elenril wonders if he should make that a redirect to http://tvtropes.org/pmwiki/pmwiki.php/Main/NietzscheWannabe
[17:47:14] <aszpain> no not server... ffmpeg command
[17:47:38] <aszpain> I get seg f in ffmpeg command the 1 time the next one i get that erro
[17:47:49] <enkidu> ffmpeg from cvs or from ubuntu repo?
[17:47:57] <kshishkov> elenril: no, it should redirect to fftropes.org
[17:48:06] <aszpain> ubuntu
[17:48:11] <enkidu> ah man...
[17:48:18] <aszpain> whats up?
[17:48:26] <enkidu> it is not up-to-date
[17:48:29] <aszpain> ubunutu 10.4
[17:48:44] <enkidu> the last one I had was cvs snapshot from 2009
[17:48:47] <aszpain> so it is a bug... and I must download from cvs?
[17:48:55] <aszpain> omg
[17:49:18] * elenril looks at himself still talking when there's Science to do
[17:49:36] <enkidu> you should: remove all libav*, ffmpeg, libx264
[17:49:43] <enkidu> install yasm
[17:49:51] <kshishkov> elenril: that's because you're not GLaD
[17:50:01] <aszpain> yasm > apt-get?
[17:50:06] <enkidu> yeah
[17:50:12] <jai> wow, looks like ffmpeg-user right about now
[17:50:13] <aszpain> more updated  things?
[17:50:25] <aszpain> ok ty
[17:50:45] * mru loads shotgun
[17:50:58] <elenril> jai: feel free to restart the git flames ;)
[17:51:05] * enkidu hides behind tree and starts coding
[17:51:48] <kshishkov> mru, you should've taked a bit of raised floor as well
[17:51:51] <elenril> or there's....release flames
[17:52:00] <elenril> btw what's holding 0.6?
[17:52:43] <siretart> elenril: I'm still waiting for mru to comment on my latest version of the symver-tricks patch
[17:52:56] <siretart> other than that, I think we're good to go
[17:52:56] <mru> siretart: I just got home
[17:53:06] <siretart> mru: hope you had a good trip!
[17:53:11] <siretart> how was LT?
[17:53:11] <mru> I did
[17:53:17] <mru> LT was good
[17:53:25] <elenril> were there trolls?
[17:53:37] <_av500_> lots
[17:53:44] <enkidu> trolls were sold to Nokia ;]
[17:53:46] <siretart> :-)
[17:54:14] <kshishkov> elenril: guess who was trolling Mozilla?
[17:54:55] <enkidu> anyways - what about, lets say, calling function each minute, that will check stream upload rate and then, if zero (or bellw threshold)
[17:55:01] <enkidu> drop stream
[17:55:19] <elenril> kshishkov: what did you do to them?
[17:55:26] <_av500_> kshishkov: is this why my firefox offered to uninstall itself today?
[17:55:41] <kierank> mine crashes every 10 mins
[17:55:47] <kshishkov> elenril: nothing, I did the same thing as on IRC - mostly napping
[17:56:01] <_av500_> ffnap!
[18:54:14] <mru> peloverde: spam watermarks?
[18:55:01] * kshishkov thinks mru is talking about new design of Nigerian passports
[18:55:33] <peloverde> Yes I'm passing judgement on the people that insist on watermarking their videos
[18:55:54] <peloverde> Nevertheless people asking how to do it are constantly asking for help
[18:55:59] <mru> I don't get it
[18:56:18] <peloverde> What's not to get?
[18:56:29] <kshishkov> they're either TV guys or greedy for fame/their movies/whatever
[18:56:46] <mru> what you're talking about
[18:56:51] <kshishkov> or just want that "Divx" logo for pretending it's encoded not with FFmpeg
[18:57:11] <mru> what overlays?
[18:57:28] <peloverde> yes
[18:57:34] <kshishkov> mru: placing small annoying picture over movie like TV logo
[18:58:08] <mru> what does that have to do with our releases?
[18:58:43] <kshishkov> there was a way to do that with hooks in FFmpeg, now it's not
[18:58:46] <peloverde> SOME users have legitamte reasons for staying with 0.5.x
[18:59:01] <peloverde> we shouldn't be telling them what useres should or shouldn't be using
[18:59:16] <mru> isn't there an avfilter gor that?
[18:59:27] <mru> for
[18:59:38] <peloverde> only in the soc repo
[18:59:56] <peloverde> and there seems to be some flamewar going on about it
[19:01:04] <mru> the situation is that diego and siretart are making releases, the distros love it, and michael and a few others hate it
[19:01:26] <mru> I will never use on of those releases, but I don't mind someone makes them
[19:01:56] <kshishkov> I think they just don't like complains like "this does not work in my release X"
[19:02:22] <mru> those who complain have failed to realase one important thing
[19:02:22] <kshishkov> if nobody bugs them about releases they won't have a thing against it
[19:02:37] <mru> distros will ship a year-old version whether it's a "release" or not
[19:02:47] <mru> with releases we at least have some control over what they ship
[19:03:15] <peloverde> agreed
[19:03:57] <mru> people building their own will continue to use svn of course
[19:04:04] <mru> they know what they're doing
[19:04:11] <peloverde> There are also still some vhook removal regressions and for some people that require vhook features not yet in libavfilter, the 0.5.x branch is their best option in the near term
[19:04:48] <siretart> more importantly, downstreams have syncronization points, so they more likely to share common copy
[19:05:00] <mru> that too
[19:06:15] * mru wonders which version http://www.youtube.com/watch?v=iIalNEW-LQ8 uses
[19:06:51] <siretart> mru: just a quick question: what is your current feeling about the current state of the symver tricks patch? - I have the impression that you don't completely object it, but I want to know :-)
[19:08:15] <mru> I don't remember which version is the latest, but I'm pretty sure we can reach a solution I'm happy with
[19:08:23] <mru> I recognise the problem, and I want it fixed
[19:09:51] <siretart> mru: my version from yesterday works on gnu, and I'm pretty confident it should do the right thing on ARM RCVT
[19:10:08] <mru> that's where we disagree
[19:10:12] <mru> but I'll fix it
[19:10:19] <siretart> ah, thanks!
[19:10:44] <siretart> I've implemented it by reading the documentation, but I don't have access to the toolchain to test myself
[19:12:58] <Dark_Shikari> KotH: no idea
[19:24:47] <iive> gcc-4.4.4 miscompiles at -O3, at least wine. coincidentally this is the level where vectorization is enabled.
[19:25:49] <kshishkov> wake me up when autovectorization in compilers works for something else than spheric horses
[19:25:53] <Dark_Shikari> try -fno-tree-vectorize
[19:28:46] <enkidu> Dark_Shikari: what was changed in ffserver, that it started to utilise cpu?
[19:29:15] <iive> i was hoping they would have fixed at least the x86 case...
[19:29:42] <Compn> enkidu : hyc made some ffserver changes
[19:30:09] <Compn> i think..
[19:30:15] <enkidu> okay. few days ago, when I was running older version, utilisation (500 connections) was 2%
[19:30:22] <enkidu> now (120 clients) is 20%
[19:30:28] <Dark_Shikari> that's what bisect is for
[19:33:06] <aszpain> When streaming.... what does rc buffer underflow mean?
[19:34:17] <Compn> when did this turn into users questions?
[19:34:22] <Compn> user questions go in #ffmpeg
[19:34:29] <aszpain> upp sorry
[19:41:26] <hyc> my changes made CPU use go *down*
[19:41:46] <hyc> it was 100% CPU when serving a live feed
[19:42:08] <enkidu> hyc, so previous changes were responsible for going UP
[19:42:50] <hyc> in that case, you need to describe which revisions you're comparing
[19:42:55] <enkidu> ffmpeg_0.5+svn20090706
[19:42:56] <hyc> "older" doesn't say much
[19:43:10] <enkidu> nearly one year old
[19:44:20] <hyc> no idea. I didn't even start working on ffserver until last month.
[19:45:06] <enkidu> ok, so we will have to investigate
[20:08:55] <Compn> heh
[20:10:37] <hyc> enkidu: what media are you serving? a static file?
[20:12:16] <enkidu> hyc: live feed
[20:13:02] <hyc> strange that you didn't have that cpu usage problem. I guess it must have been a regression from the version you ran
[21:13:36] <Yuvie> gah! the vp8 spec only defines sub version 0 as valid, but the test vectors use sub version 1 and 2 as well!
[21:14:09] <Dark_Shikari> it'd be easier if you just ignored the spec
[21:15:01] <Yuvie> I actually stopped relying on the spec weeks ago
[21:15:40] <Yuvie> just annoying since I thought I could ignore the bilinear filter and other things version 1 and 2 imply
[21:21:35] <Yuvie> other fun fact: if a frame says to update the golden reference with the current altref and the altref with the current golden reference, libvpx copies the golden into the altref and then back into the golden
[21:22:09] <mru> lol
[21:23:42] <Dark_Shikari> is that a bug?
[21:24:01] <Dark_Shikari> or just a waste of time
[21:24:08] <Dark_Shikari> and wait, why would the stream say such a thing?
[21:24:22] <Yuvie> no reason for it to, but it can
[21:24:25] <Yuvie> so just a waste of time really
[21:24:56] <Yuvie> (yes, libvpx does full-frame copies into alt/golden references instead of using a pointer)
[21:25:07] <Dark_Shikari> lol
[21:25:17] <Dark_Shikari> I guess that's needed if you need to filter each one differently?
[21:25:46] <Yuvie> why would you filter each differently?
[21:26:01] <Dark_Shikari> there are filters you can apply to ref frames iirc
[21:26:03] <Dark_Shikari> i.e. denoise, etc
[21:26:11] <Dark_Shikari> or sharpen
[21:26:17] <Dark_Shikari> Or maybe that's only for encoding-side on the altref?
[21:26:52] <Yuvie> there's nothing in the bitsteam to signal that
[21:27:23] <Dark_Shikari> ah k
[21:27:26] <Dark_Shikari> so it must be only encoder-side
[21:45:18] * lu_zero -> home!
[21:48:53] <KotH> Dark_Shikari: well, could you check?
[21:49:06] <KotH> Dark_Shikari: you were the one who debugged together with that guy in #mplayer :)
[22:00:24] <KotH> night girls
[23:00:28] <Kovensky> KotH: wishful thinking?
[23:58:00] <barque> well hello again!
[23:58:16] <barque> kshishkov, dobre den/noche, kaktila?
[23:59:24] <iive> barque: he usually goes to sleep quite early, so he may not answer you until tomorrow


More information about the FFmpeg-devel-irc mailing list