IRC Logs

*******************************************************************
[03:26] phlog ([email protected]) joined #gstreamer.
[03:27] <phlog> does anyone know if anyone's working on an mplayer-like command-line video/audio player based on gstreamer?
[03:28] <walters> gst-launch :)
[03:28] <hyriand> if you don't need a gazillion options, it should be easy enough to make a wrapper script around gst-launch..
[03:29] <phlog> cool
[03:31] <phlog> I'll look into it
[03:38] Zeenix ([email protected]) left irc: "The whole problem with the world is that fools and fanatics are always so certain of themselves, but wiser people so full of d"
[03:39] jcsston` ([email protected]) joined #gstreamer.
[03:48] pb_ ([email protected]) left irc: "Client exiting"
[03:49] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[03:55] Nick change: jcsston` -> jcsston
[04:31] sublett ([email protected]) left irc: "I like food, food is good!"
[04:37] phlog ([email protected]) left #gstreamer.
[05:31] djileum ([email protected]) joined #gstreamer.
[05:32] <djileum> hello
[05:38] aeyakovenko ([email protected]) joined #gstreamer.
[05:38] <aeyakovenko> is there still a pluginstamp.sh to help with plugin development?
[05:54] <djileum> does anyone know the best way to get the "mad" element using the default rpms for redora...I'm trying to get rhythmbox installed and it's just missing "mad"
[05:55] aeyakovenko ([email protected]) left #gstreamer.
[05:58] <djileum> i'm trying to install gstreamer-plugins-extras/mp3 from freshrpms.net
[06:01] <djileum> and it tells me "osscommon: unable to open /dev/dsp (in use?)" 5 times
[06:01] thaytoo ([email protected]) left irc: "leaving"
[06:01] <djileum> i'd appreciate any help
[06:02] <taf2> djileum: yes
[06:03] <djileum> taf2...what am I missing (rpm, etc..) that should have been installed?
[06:03] <djileum> i followed directions on a redhat board
[06:03] <djileum> to install those 2 rpms off of freshrpms.net, and now i get /dev/dsp problems
[06:03] <taf2> djileum: the fact that it says /dev/dsp is in use that's tricky
[06:04] <djileum> i've got gaim running....does that have naything to do with this?
[06:04] <djileum> never mind..i just quit gaim and still got it
[06:05] <taf2> when i said "yes" i was took quick to reply not having finished reading to see your problem ;)
[06:05] <taf2> i have sometime though so i'll help you look for solution
[06:05] <djileum> :)...i've got esound as well apparently (that's what xmms uses)
[06:05] <djileum> great!!! thank you so much
[06:05] <djileum> i've spent hours on this (literally)
[06:08] <taf2> djileum: http://www.guerrilla.net/reference/dsp/prog_dsp.htm
[06:08] jaitly ([email protected]) joined #gstreamer.
[06:08] <taf2> gives some c code that opens /dev/dsp
[06:08] <taf2> does /dev/dsp exist on your system?
[06:09] <djileum> yes
[06:09] <taf2> mine looks like this: crw-------    1 taf2     root
[06:09] <djileum> xmms, xine, etc...
[06:09] <djileum> everytyhing works fine
[06:09] <djileum> crw-------    1 djileum  root      14,   3 Sep 15 08:40 dsp
[06:10] <taf2> maybe a process is running using that device? 
[06:10] Action: taf2 i'm just guessing...
[06:10] aeyakovenko ([email protected]) joined #gstreamer.
[06:10] <djileum> hmm
[06:10] <aeyakovenko> is there an uptodate howto on building a new element?
[06:10] <djileum> i don't think so...
[06:10] <djileum> i just did ps ax
[06:10] <djileum> couldn't see anything
[06:10] <taf2> k
[06:11] <djileum> i *did* see [play <defunct>]
[06:11] <taf2> that might it :)
[06:11] <taf2> i was going to say the next thing is to figure out where osscommon error is coming from
[06:11] Action: djileum is soooo tempted to say "screw this!" and go back to using xine
[06:11] <taf2> djileum: hate to say it but maybe a reboot will fix it
[06:12] <djileum> really...
[06:12] <djileum> hmm
[06:12] <djileum> i'll brb then :)
[06:12] <taf2> if its a defunct process
[06:12] <taf2> k
[06:12] djileum ([email protected]) left irc: Remote closed the connection
[06:17] djileum ([email protected]) joined #gstreamer.
[06:17] <djileum> well a reboot didn't fix it sadly
[06:19] <djileum> i'm not sure why /dev/dsp is being monopolized
[06:19] <djileum> will installing gstreamer-esound work?
[06:19] <aeyakovenko> so are there any plugin docs?
[06:20] <aeyakovenko> like a howto
[06:21] <djileum> [play <defunct>] no longer appears...but 
[06:21] <djileum> yeah
[06:24] <taf2> djileum: damn 
[06:24] <taf2> djileum; you might want to try the fedora list for this
[06:24] <taf2> sounds like an rpm issue to me
[06:24] <taf2> cause on my fedora system the rpms you tried just worked
[06:25] <djileum> odd
[06:26] <djileum> this is what I have installed:
[06:26] <djileum> gstreamer-plugins, extras , plugins-mp3, tools and the actual gstreamer rpm itself...all version 0.6.3, and plugins,extrsa from freshrpms
[06:27] <djileum> and of course, rhythmbox 0.6.5-0.1.fr (From freshrpms)
[06:27] <djileum> is the rhythmbox that i'm using the problem?
[06:27] <djileum> perhaps use a non-freshrpms version? (like hte version that came with fedora core 1?)
[06:28] <taf2> djileum: that might be it
[06:28] <djileum> okay
[06:28] <djileum> you wouldn't know offhand which cd rhythmbox is on would you?
[06:28] <taf2> i don't but it might be easier to just grab it from yum or apt ;)
[06:29] <djileum> ah
[06:29] <walters> the version in fc1 is ancient
[06:29] <djileum> ok
[06:33] <djileum> mann...i'm about ready to cry
[06:35] ChriHJW_log ([email protected]) left irc: Connection timed out
[06:37] <djileum> i need some piece of hardware that's old so i can beat it to death
[06:39] <djileum> for some reason...it works for everyone but me
[06:40] <djileum> now..the version I have installed is 0.6.3.  I discovered that gnomevfssrc isn't in the fc1 packages for gstreamer-0.6.4...did someone fix this?
[06:43] Action: djileum is going back to rhythmbox/xine && xmms for the time being i guess
[06:44] <djileum> thanks for the help anways taf2
[06:45] Action: djileum is outta here
[06:45] <djileum> night all
[06:45] djileum ([email protected]) left irc: "Leaving"
[06:46] <aeyakovenko> is there a helloworld gst plugin?
[06:52] thaytan ([email protected]) joined #gstreamer.
[07:28] Nick change: mathrick|sleep -> mathrick
[07:28] <mathrick> mornin' people
[07:34] somex1 ([email protected]) joined #gstreamer.
[07:55] Nick change: mathrick -> mathrick|bath
[07:57] ChrisHJW ([email protected]) joined #gstreamer.
[07:59] bluejay ([email protected]) left irc: Remote closed the connection
[08:19] aeyakovenko ([email protected]) left #gstreamer.
[08:20] ChrisHJW_log ([email protected]) joined #gstreamer.
[08:20] ChrisHJW ([email protected]) left irc: Read error: 54 (Connection reset by peer)
[08:22] bluejay ([email protected]) joined #gstreamer.
[08:26] md` ([email protected]) left irc: Read error: 110 (Connection timed out)
[08:26] bluejay ([email protected]) left irc: Client Quit
[08:45] somex1 ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[08:49] Nick change: mathrick|bath -> mathrick
[09:14] kmaraas ([email protected]) left irc: "Leaving"
[09:16] <mathrick> bbl
[09:16] mathrick ([email protected]) left irc: "Client exiting"
[09:27] Rotty ([email protected]) joined #gstreamer.
[09:29] thaytan ([email protected]) left irc: Read error: 110 (Connection timed out)
[09:31] md` ([email protected]) joined #gstreamer.
[09:41] markey ([email protected]) joined #gstreamer.
[09:54] thomasvz ([email protected]) left irc: Read error: 110 (Connection timed out)
[10:00] jaitly ([email protected]) left irc: "Leaving"
[10:11] thomasvs ([email protected]) joined #gstreamer.
[10:19] KoRnouille ([email protected]) joined #gstreamer.
[10:38] Kaetzchen ([email protected]) joined #gstreamer.
[10:39] alley_cat ([email protected]) left irc: Nick collision from services.
[10:39] Nick change: Kaetzchen -> alley_cat
[10:39] <markey> nice nick, Kaetzchen 
[10:39] <markey> ;)
[10:39] <alley_cat> :)
[10:41] thaytan ([email protected]) joined #gstreamer.
[10:56] Shoragan ([email protected]) joined #gstreamer.
[11:02] dolphy ([email protected]) joined #gstreamer.
[11:03] <thaytan> rah, #gst
[11:38] thomas_ ([email protected]) joined #gstreamer.
[11:38] thomas_ ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[11:44] ensonic ([email protected]) joined #gstreamer.
[11:44] <ensonic> hi hi
[11:44] <thaytan> yo
[11:44] <md`> thaytan: [11:43] #gst unable to join channel (too many channels open)
[11:44] <md`> :P
[11:54] <ensonic> I am trying to build gst-plugins rom cvs now, when running autogen.sh it abort with :/usr/local/bin/m4: Error matching regular expression ....
[11:55] <ensonic> m4 is GNU m4 1.4
[11:55] <ensonic> automake is 1.7
[11:56] <thaytan> does it get anywhere at all?
[11:57] <ensonic> gstreamer itself build and installed okay
[11:57] <ensonic> I try updating automake to 1.8
[11:57] <thomasvs> ensonic: why do you have m4 in /usr/local ?
[11:58] <ensonic> thomasvs: why not (it's a sun and I have replaced their old m4)
[11:58] <thomasvs> ensonic: I'm just wondering, because as soon as people do all sorts of customization things tend to go wrong :)
[11:59] <thomasvs> ensonic: in any case, we'll need more output to make a guess at what's wrong
[12:00] <ensonic> thomasvs: the path of the m4 binary is not the problem, look more like the is a bug in automake 1.7, which version of autoconf&automake you are using?
[12:00] jdahlin ([email protected]) joined #gstreamer.
[12:02] <thaytan> thomasvs: hey, looks like we'll be in Barcelona 21,22 & 23rd April and maybe the night of the 20th late. Sound ok?
[12:03] sublett ([email protected]) joined #gstreamer.
[12:03] <thomasvs> ensonic: 2.57 & 1.7.8
[12:04] <thomasvs> thaytan: yeah, sounds ok
[12:04] <thomasvs> thaytan: that means, I don't have anything planned that far ahead :)
[12:05] <thaytan> heh. we can always find a hotel if your plans change 
[12:05] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[12:06] <thomasvs> well, you're noted in the calendar, so I'm sure I'll remember
[12:11] <thaytan> ta
[12:11] <thaytan> it's going to be great fun, I wish we had longer!
[12:13] <ensonic> on http://www.freedesktop.org/~gstreamer/apps/ideas.html sould be a link to  mixer.png (this seems to got lost during transfer of the homepage)
[12:31] Company ([email protected]) joined #gstreamer.
[12:41] jdahlin ([email protected]) left irc: "Leaving"
[12:54] walters ([email protected]) left irc: "out"
[12:54] jdahlin ([email protected]) joined #gstreamer.
[12:56] Nick change: markey -> markey|afk
[13:14] gheet ([email protected]) joined #gstreamer.
[13:20] <ensonic> it was a bug in autoconf, now autogen runs through
[13:20] <ensonic> erm automake was the buggy one
[13:33] Company ([email protected]) left irc: Remote closed the connection
[13:41] Nick change: KA_afk -> KA
[13:54] Company ([email protected]) joined #gstreamer.
[14:19] Company ([email protected]) left irc: Remote closed the connection
[14:42] sxpert ([email protected]) left irc: Read error: 110 (Connection timed out)
[14:43] LeRoutier ([email protected]) joined #gstreamer.
[14:44] <LeRoutier> hello
[14:44] <LeRoutier> CIA is dead, that's sad
[14:48] Uraeus ([email protected]) joined #gstreamer.
[14:48] <jdahlin> lo Uraeus
[14:48] <Uraeus> scooby doo
[14:49] <Uraeus> jdahlin: so have you completed rewriting the new scheduler yet? ;)
[14:49] <jdahlin> Uraeus, well well well :)
[14:50] <jdahlin> let's take it easy in the beginning, shall we?
[14:50] <Uraeus> what? are you slacking on the first day at work :)
[14:51] Action: jdahlin opens up gst/schedulers/gstoptionalscheduler.c
[14:53] <Uraeus> heh :)
[14:58] <Uraeus> jdahlin: I hope you fix cothreads on i686 before you leave the office today then :)
[15:04] Nick change: markey|afk -> markey
[15:14] AlexCONRAD ([email protected]) joined #gstreamer.
[15:14] KoRnouille ([email protected]) left irc: Read error: 60 (Operation timed out)
[15:18] sxpert ([email protected]) joined #gstreamer.
[15:24] <Uraeus> hmm, I am still have tons of mp3's which with current CVS fail with: RUNNING pipeline ...
[15:24] <Uraeus> ERROR         filesrc( 4852) gstfilesrc.c(750):gst_filesrc_open_file: <filesrc0> (null)
[15:24] <Uraeus> ERROR: from element /pipeline0/filesrc0: Resource not found.
[15:24] <Uraeus> ERROR: pipeline doesn't want to play.
[15:25] <Uraeus> err forget it
[15:26] <Uraeus> strange but that line works when I give the file a more easily parsed name, but in rhythmbox it just gets 'skipped' over
[15:28] <ensonic> grmf, I've already submitted 3 new fixes for compiling gst-plugins cvs :-(
[15:29] <Uraeus> ensonic: that is the price for running weird setups (on the other side we are happy that you are and are making patches that improve general portability :)
[15:30] <ensonic> Uraeus: he a double ';;' in a variable declaration is simply no valid C and should not be commited 
[15:30] <Uraeus> ensonic: its a gcc'ism?
[15:30] <ensonic> Uraeus: is what? I am using gcc
[15:31] <bitshifter> I think the problem is with gcc-2.95 and all compilers that don't allow variables to be declared in the middle of a block. The second ';' is understood as an empty statement, so after that all variable declarations are considered to be 'in the middle of a block'
[15:33] <ensonic> this is c, only c++ allows to mix declarations and statements
[15:33] <ensonic> I pesonally like the c++,java way of mixing them, but if one chooses c ...
[15:33] <bitshifter> I thought C99 does as well 
[15:34] <bitshifter> *shrug*  I just know it's something gcc-2.9x is picky about
[15:34] <ensonic> I wont bet about it, but anyway in the current case it's a type so it should be fixes
[15:36] <ensonic> other things that I've found are missing includes and I've check witha linux box, that the same includes are used [bzero()]
[15:38] <LeRoutier> ensonic, you're under which OS ? solaris ? a BSD ?
[15:38] <Uraeus> jdahlin: it mattias in the office?
[15:38] <ensonic> the worst thing is that the sun is slow in compiling (300 Mhz) so it takes ages ...
[15:38] <jdahlin> Uraeus, AFAICS
[15:38] <Uraeus> jdahlin: tell him that the Fedora development apt repo doesn't seem to work fully atm
[15:39] <ensonic> LeRoutier: Sun?Sparc Solaris 2.6 - I know its not a common environment, anyway the whole GNU stuff is there and quite up-to-date
[15:39] <jdahlin> Uraeus, poke him on im?
[15:39] <Uraeus> jdahlin: where?
[15:40] <jdahlin> Uraeus, no idea :)
[15:40] <Uraeus> jdahlin: doesn't he sit in the office where you are?
[15:40] <jdahlin> Uraeus, he seems to be pretty busy doing something
[15:41] <ensonic> what is the procedure to get cvs-write access (at least for such simple fixes, this is easier that filing a report in bugzilla)
[15:42] <Uraeus> jdahlin: chicken :)
[15:43] <jdahlin> Uraeus, tss
[15:43] Thias ([email protected]) joined #gstreamer.
[15:44] <jdahlin> Uraeus, there you go
[15:44] <Uraeus> hi Thias
[15:44] <Thias> yeah ;-)
[15:44] <Thias> hi there Uraeus!
[15:44] <Uraeus> Thias: I am not able to apt-get the Fedora development packages
[15:44] <Thias> s/ayo.freshrpms.net/ayo.us.freshrpms.net/
[15:44] <Uraeus> ah, ok thanks :)
[15:45] <thomasvs> Uraeus: let the poor man work, you slavedriver
[15:45] <Thias> the main download server gets out of synch with the devel tree quite often... I don't mirror the main Red Hat and Fedora stuff there, so I can't do much about it
[15:45] <Thias> thomasvs ;-p
[15:46] <Uraeus> thomasvs: hey just because I beat on the drums doesn't mean I am less of a slave the rest of the people on the galley :)
[15:48] <Uraeus> Thias: ok, my update works now. You and jdahlin can go back to work now ;)
[15:49] Nick change: KA -> KA_afk
[15:53] <Thias> who ever said I was working? :-D
[16:00] <Uraeus> Thias: btw, now that you have both thomasvs and jdahlin just idling in the office, make sure to get one of them to fix so that GStreamer works on your machine
[16:03] <Uraeus> Thias: that way freshrpms can become the primary resource for updated rpms after each release and we don't have to tell the angry masses that they have to wait 6 months on feodra.us QA to complete :)
[16:03] <thomasvs> Uraeus: so you're using matthias as an excuse for your laziness ?
[16:03] <Uraeus> thomasvs: no, I am focusing on the solution, not the problem :)
[16:04] <Uraeus> thomasvs: and since fedora is a problem, I am pointing towards freshrpms as the solution
[16:05] <thomasvs> Uraeus: the solution is you get off your lazy ass like we agreed :)
[16:06] <Uraeus> thomasvs: my cute ass isn't the problem, the problem is that even with my ass solidly plantet in fedora it would still take 1-2 more asess in addition to mine and yours to get something done there
[16:07] <thomasvs> Uraeus: you probably didn't listen to what we discussed last time.  It was fairly simple.
[16:07] <thomasvs> a) we make packages that are candidates for fedora and put them online on gst as well
[16:08] <thomasvs> b) we work to get them polished and put in fedora.us repos
[16:08] <thomasvs> if you were to help out on fedora.us, and follow the procedure, each package would be accepted very quickly.
[16:08] <thomasvs> but even if you don't, they're still available on our site
[16:08] <Uraeus> thomasvs: no, cause it takes more than one reviewer to review those packages
[16:09] <thomasvs> Uraeus: yep, so you are one of the persons needed, and there are other people I know too
[16:09] <Uraeus> thomasvs: and why should we spend a lot on energy on making fedora packages when freshrpms can provide them?
[16:09] <thomasvs> Uraeus: because we cannot submit packages to freshrpms
[16:09] <thomasvs> Uraeus: why are we having the same discussion all over again for the fifth time ?
[16:09] Action: thomasvs goes to do something useful instead
[16:09] <Uraeus> thomasvs: we don't need to, I am sure mathias would be happy to make any packages we ask if things work for him instead
[16:10] Action: Uraeus grumbles
[16:10] <thomasvs> Uraeus: I guess you don't understand, which is fine, but no need to discuss it with me anymore
[16:10] <thomasvs> Uraeus: you've already made it clear you do not want to help, and that is fine with me.
[16:11] Action: Uraeus points out that it was thomasvs who started discussing it with him not the other way around
[16:16] <thomasvs> no, I was just saying that you helping out would have been a perfectly viable solution either :)
[16:16] <thomasvs> Then I clarified when you disagreed :)
[16:17] <Uraeus> ok, lets leave it at that before fedora grows into a real issue between us :)
[16:18] <ensonic> thomasvs: (repeat of question from 15:41) what is the procedure to get cvs-write access (at least for such simple fixes, this is easier that filing a report in bugzilla)
[16:23] smoke ([email protected]) joined #gstreamer.
[16:23] <smoke> is there something wrong configured or why does gst-player have problems with m3u files?
[16:24] <LeRoutier> m3u aren't supported, they are plain XML files that only contains metadata + URIs if i remember well (or i mix it with .ASX)
[16:24] <smoke> xml files?
[16:25] <smoke> m3u is simply a file like http://serverurl/mountpoint
[16:25] <smoke> its should really be implemented
[16:25] <LeRoutier> so i did a mix with .asx files in my head
[16:26] <smoke> well xmms is better for playing stuff anyway ...
[16:27] <smoke> hmm :/
[16:27] <smoke> rhythmbox sucks on playing live streams also
[16:27] <smoke> couldnt find source pad ...
[16:28] <smoke> i allways thought gstreamer was some kind of stable but this is just plain wrong as i just realize
[16:28] <Uraeus> smoke: woke up on the wrong side of bed today? are you using 0.7?
[16:29] <smoke> hmm 0.6.4 says gst-properties
[16:30] <smoke> well i just thought playing an mp3/ogg stream would be painless
[16:30] <Uraeus> hmm, the live streams worked well for me the last time I used rb and the 0.6 branch
[16:31] <Thias> I get a connection refused error for www.gstreamer.org :-(
[16:31] <Uraeus> Thias: its gstreamer.net (or is the 'new' domain)
[16:31] <smoke> yeah some time ago it was working i just dont know what went wrong
[16:31] <Uraeus> s/or/org/
[16:31] <Thias> oops, indeed
[16:32] <Thias> I just went to gstreamer.sf.net anyway ;-)
[16:32] <dolphy> many people from polar.es3.egwn.net here :)
[16:32] <smoke> i think maybe debian package problem
[16:33] <Uraeus> dolphy: yeah, you are doing well at growing the channel :)
[16:33] <smoke> maybe apt-get resolves the problem
[16:34] <smoke> looks like theres a new rhythmbox package
[16:34] <dolphy> Uraeus: see ! :)
[16:34] <jdahlin> I have a problem with gst-register
[16:34] <jdahlin> it can't seem to find gstmikmod, even though it's built and installed
[16:35] <smoke> hmm hangs really ugly on opening a http:// url
[16:37] <Uraeus> smoke: check with walters when he pops in, he maintains RB and is a debianite
[16:38] <Uraeus> err actually he just got hired by Red Hat so the last part is not true anymore :)
[16:47] <dilinger> Uraeus: well, he's still a debianite.  at least, he's accepted patches for cdbs.
[16:47] <Uraeus> dilinger: ok, well I know he did install Red Hat on his machine the day after he got the job or something like that, but maybe he kept a debian partition around
[16:48] <smoke> heh suprnova.org doesnt exist anymore, they now made a commercial product out of it??
[16:49] <LeRoutier> the torrent site  ?
[16:49] Nick change: Uraeus -> Ura_gym
[16:50] <smoke> LeRoutier: yes
[16:50] yippi ([email protected]) joined #gstreamer.
[16:50] <smoke> when i check suprnova.com i must nearly vomit
[16:51] <LeRoutier> 100% Legal Software! ??? hum, i don't know why but i don't really trust this sentence
[16:52] Company ([email protected]) joined #gstreamer.
[16:52] <smoke> hehe yes
[16:52] <smoke> they are just idiots... imho i dont think it has anything to do with suprnova.org at least i hope so
[16:53] <thomasvs> Company: what is the correct way to get a mime type from caps ?
[16:54] <ensonic> in cvs under gst-plugins is a module gst/tcp, 1.) this has no configure option to disable building it, 2.) it uses inet_aton without checking for it (it is not available on most unix boxes)
[16:54] <ensonic> any ideas how I can continue trying to build the gst-plugins package?
[16:54] <LeRoutier> smoke, look at the bottom of the page, happy search engine
[16:55] <thomasvs> ensonic: add a check for inet_aton ?
[16:56] <LeRoutier> solaris 2.6 missing IPv6 aware inet fcts ?
[16:57] <Company> thomasvs: mime type from caps?
[16:57] <Company> thomasvs: you mean the audio/x-raw-int string?
[16:57] <Company> thomasvs: gst_structure_get_name
[16:58] <ensonic> thomasvs: i've got already a good bunch of patches submitted to bugzilla, I've just copied a implementation from a different package to see if the cvs release works at all
[16:58] <bitshifter> that needs fixing in the App Devel handbook under 'pads' btw
[16:58] <thomasvs> Company: argh, that's why I missed it.  any reason why it's not called _mime ?
[16:59] <thomasvs> bitshifter: yep, agreed. can you file a bug so we don't forget ?
[16:59] <Company> thomasvs: because a gststructure is a general thing and not caps specific
[16:59] <bitshifter> section 13.2
[17:00] <thomasvs> Company: http://thomas.apestaart.org/download/screenshots/gst-player-error-type.png
[17:01] <thomasvs> Company: what I want to do is provide this ...
[17:01] <thomasvs> Company: by checking agains the list of exported mime types that the codecs handle
[17:01] <bitshifter> sure
[17:01] <thomasvs> Company: where should I best do this in spider ?
[17:01] Nick change: AlexCONRAD -> KoRnouille
[17:03] <Company> thomasvs: you want to do what?
[17:05] <thomasvs> Company: ok, the goal is to have apps give an error when there is no decoder for the given file
[17:05] <thomasvs> Company: so, during autoplugging, through typefind spider finds out the mime type, right ?
[17:05] jcsston ([email protected]) joined #gstreamer.
[17:05] <thomasvs> (for at least a bunch of files)
[17:06] <thomasvs> some of those we won't have a decoder for, ie no element that has that mime type in it's sink pad template
[17:06] <thomasvs> so, in that case spider can error out saying "there's no decoder installed"
[17:06] <thomasvs> which would be the dialog box you see
[17:06] <Company> ah
[17:06] <thomasvs> I did this for now in the path plug function, where it returns LINK_REFUSED, but htat's of course not the right way to do it
[17:06] <Company> spider doesn't do that unfortunately
[17:06] <thomasvs> no, but I was assuming it wouldn't be too hard to add
[17:07] <thomasvs> after typefinding, just check if the mime type it found is somewhere in the templates ?
[17:07] <Company> no idea
[17:07] <Company> i'm not going to add anything to spider ;)
[17:07] <thomasvs> heh, agreed :)
[17:07] <Company> the new autoplugger will do that of course
[17:07] <thomasvs> but right now when you play a file it can't play, all sorts of things can go wrong
[17:07] <thomasvs> so as a first fix, checking the type found against the plugins should be an easy first step
[17:08] <thomasvs> all I want to know is, where in the spider source does it know the mime type of the input stream ?
[17:12] <Company> gst_spider_identity_plug is the first function that has that knowledge
[17:13] <Company> you'd need to follow the function it calls to find the best place for that error though
[17:16] Action: Company doesn't get the code himself anymore ;)
[17:17] <thomasvs> I don't blame you for not wanting to work on spider :) I hope you guys figure out how to do the new autoplugger
[17:19] <Company> what should happen when you have an AVI that has an unknown video codec inside btw?
[17:20] <thomasvs> it would throw the same error, but later down the plugging
[17:20] <thomasvs> does spider have a concept of "i've tried plugging everything I can think of, I give up" ?
[17:20] <Company> no
[17:20] <thomasvs> basically, I want us to be able to throw any file at it, and do something sensible with it, even if the file isn't :)
[17:20] <Company> because demuxers don't have a concept of "i've thrown out all pads i can"
[17:21] <Company> there's a problem though
[17:21] <Company> the autoplugger shouldn't error out if you only want the audio part
[17:21] Action: ensonic is doing the patches for the configure stuff the the author of the tcp plugin should have done :-(
[17:21] <Company> so plugging an avi with unknown video from rhythmbox shouldn'T throw an error
[17:26] <thomasvs> hm, yeah
[17:26] <thomasvs> or no, but yeah :)
[17:26] <thomasvs> I know what you mean
[17:27] <thomasvs> in any case, if typefind finds a type, but we don't have a decoder/demuxer for it, then it's best to throw an error
[17:27] <Company> my plan was to never throw an error in that case
[17:27] <thomasvs> as for the rhythmbox case - some people would maybe consider it an error
[17:27] <thomasvs> so we could have spider have a property for that, maybe
[17:27] <Company> but have a custom function for the autoplugger for "could not connect to sinkpad X"
[17:28] <Company> s/function/signal/
[17:28] <Company> srcpad X
[17:28] <Company> so we'd emit a no-connection signal on the video when someone plays an mp3
[17:29] <Company> and the same if we don't find a video codec for AVI
[17:30] <Company> there might be ause for a signal when we have a pad we can't plug
[17:30] <Company> worst case is ogg btw
[17:31] <Company> maybe the ogg contains video, maybe not
[17:31] smoke ([email protected]) left irc: Remote closed the connection
[17:33] <Company> asf is the same
[17:35] trow ([email protected]) joined #gstreamer.
[17:57] jcsston ([email protected]) left irc: Read error: 60 (Operation timed out)
[18:04] <Company> thomasvs: http://freedesktop.org/cgi-bin/viewcvs.cgi/gstreamer/gstreamer/gst/gstelement.h.diff?r1=1.161&r2=1.162 - was that on purpose?
[18:11] <thomasvs> ah, no, I wanted to ask you about that, must have commited by accident
[18:12] Nick change: harshy -> harshyWork
[18:12] <thomasvs> Company: why is that there, it gives output on stdout, when it laready throws the error
[18:12] <Company> thomasvs: it gives a debugging output message - on cvs builds error messages are always enabled
[18:13] <Company> thomasvs: it causes no output on releases (if it does it's a bug)
[18:13] <Company> thomasvs: and it's there because not all apps print errors
[18:14] <Company> thomasvs: and it's a good idea to include errors in the debugging system
[18:17] <thomasvs> yeah, they were in the debugging system if you asked for it - I just don't like it being printed always since you already get the feedback in the usual way
[18:17] <taaz> thomasvs: did you get my mail with the pkconfig patch?
[18:17] <thomasvs> anyways, if you think it really is necessary, put it back in
[18:17] <thomasvs> taaz: yeah, hope to look at it today
[18:17] Nick change: KA_afk -> KA
[18:17] <thomasvs> Company: but it would have been nice if you had let me know you wanted that in, then I wouldn't have overlooked it
[18:18] <Company> thomasvs: we had talked about that before
[18:18] <Company> thomasvs: anyway, i'll put it back in
[18:24] <thomasvs> no, as a reply to my RFC mails
[18:25] <taaz> thomasvs: ok.
[18:27] <jdahlin> gst-md5sum-0.7 filesrc location="../medium/alien.mpg" ! mpegdemux name=demux ! { queue ! fakesink } { demux. ! queue ! md5sum }
[18:27] <jdahlin> gst-md5sum-0.7 filesrc location="../medium/alien.mpg" ! mpegdemux name=demux ! { queue ! fakesink } { queue ! md5sum }
[18:27] <jdahlin> Why do these two command lines produce different md5sums?
[18:27] <taaz> what should i do with plugins glib-mkenums patch?  would be nice to see it in the next release since python non-core libs bindings will require it.
[18:28] <jdahlin> eg, what does demux. ! do ?
[18:28] <Company> jdahlin: because the second line doesn't link queue to anything
[18:28] <Company> jdahlin: things with a dot are a reference
[18:28] <thomasvs> taaz: what's the equivalent of that in the core ?
[18:28] <jdahlin> Company, ok
[18:28] <Company> bla.blubb refereces the element named "bla" with the pad named "blubb"
[18:29] <jdahlin> Company, and should { demux. ! queue ! mad ! md5sink } take like forever to run?
[18:29] <Company> bla. references any pad of element "bla"
[18:29] <taaz> thomasvs: gst/Makefile.am does some stuff with glib-mkenums to dump out type stuff for various enums
[18:29] <Company> .blubb references last element, pad named "blubb"
[18:30] <taaz> only thing i'm wondering about is the probably-never-tested configure flags in core to shut off mkenums code
[18:30] <Company> jdahlin: i don't think it should run forever - nothing should run forever as long as the input is finite ;)
[18:30] <jdahlin> Company, it seems to be hung doing nothing
[18:31] <Company> probably an EOS issue somewhere
[18:31] <jdahlin> poll([{fd=4, events=POLLIN}], 1, 0)     = 0
[18:31] <jdahlin> gettimeofday({1075742543, 82305}, NULL) = 0
[18:31] <jdahlin> etc
[18:31] <jdahlin> complete cmd line: $ gst-md5sum-0.7 filesrc location="../medium/alien.mpg" ! mpegdemux name=demux ! { queue ! fakesink } { demux. ! queue ! mad ! md5sink }
[18:32] somex1 ([email protected]) joined #gstreamer.
[18:34] Action: Company wonders why such hacky tools like gst-md5sum even get installed
[18:34] <Company> anyway, no idea which element didn't hit EOS and why
[18:34] <jdahlin> it's good for testing
[18:36] <Company> yeah, but not good for installing on anybodys computer
[18:36] <Company> anyway, i think it's because the audio part is connected to the fakesink
[18:37] <Company> $ gst-md5sum-0.7 filesrc location="../medium/alien.mpg" ! mpegdemux name=demux ! { queue ! mad ! md5sink }
[18:37] <Company> that should work
[18:37] <jdahlin> hmm. how do I know which part is the audio and which is the video?
[18:37] <Company> you can name the pads explicitly
[18:37] <Company> $ gst-md5sum-0.7 filesrc location="../medium/alien.mpg" ! mpegdemux name=demux .audio_00 ! mad ! md5sink
[18:38] <Company> or when using mad, y<ou can just omit it, because it won't connect to the video sink anyway
[18:38] <jdahlin> why audio_00?
[18:38] <Company> it's the name of the pad you want to connect to
[18:38] <jdahlin> first audio stream?
[18:38] <Company> gst-inspect shows you the names
[18:38] <jdahlin> and there's only 100 possible in mpeg?
[18:39] ChrisHJW ([email protected]) joined #gstreamer.
[18:39] <Company> SRC template: 'audio_%02d'
[18:39] <Company> so the 101st is audio_101 :)
[18:39] <jdahlin> hackish!
[18:40] <Company> you'Re free to solve it in a better way for 0.9 :p
[18:40] <jdahlin> maybe I will
[18:40] <jdahlin> anyway
[18:40] <Company> in fact i'd like it if someone invented a better way for that
[18:40] <jdahlin> to check the md5sum of the video part?
[18:40] <Company> look at gst-inspect output
[18:40] <jdahlin> mpegdemux ! { queue ! md5sink } is not correct I suppose
[18:41] <Company> it tells that video parts are named video_%02d
[18:41] <Company> which means the first video pad is named video_00
[18:41] <jdahlin> I can't see anything related to video from the output of gst-inspect
[18:42] <Company> near the top
[18:42] <Company> Pad Templates:
[18:42] <Company> after that comes the sink template
[18:42] <Company> and after that comes the video template
[18:42] <jdahlin> nope
[18:43] <jdahlin> how should the command line look like?
[18:43] <Company> gst-isnpoect-0.7 mpegdemux
[18:43] <Company> gst-isnpect-0.7 mpegdemux
[18:43] <jdahlin> much better, danke
[18:43] <Company> you'Re familiar with what that command outputs?
[18:44] <jdahlin> not really
[18:45] <Company> command is gst-inspect <elementname>
[18:45] <Company> the first part shows general infos about the element (d'oh)
[18:46] <Company> the second part and the most interesting normally are the pad templates
[18:47] <Company> every pad (i don't think it's enforced yet but should) depends on a pad template
[18:47] <Company> templates tell you if it's a source or sink, the availability (always, sometimes or request) and the caps
[18:49] <Company> flags, implementation, clocking and indexing are info about scheduler and clock interaction and other "how it works" stuff
[18:49] <Company> the pads just lists which pads are available by default
[18:50] <Company> the properties are interesting again, because you can set the via gst-launch :)
[18:53] <Company> signals are just the signals the element may emit
[18:53] <Company> and dynparams are dynamic parameters, which are nowadays mostly unused
[18:54] <Company> i don't even know myself how they work
[18:54] <Company> they're supposed to smartly adapt property changes somehow so you don't get cracks when setting volume from 1 to 100 with a click
[18:56] <Company> ok, this concludes our short intro to gst-inspect - if there are questions left, don't hesitate to ask. We wish you a pleasant stay in this channel
[18:57] yippi ([email protected]) left irc: "Client exiting"
[18:58] Action: thomasvs stops filming now
[19:00] Action: Company waits for Uraes to scream "this needs to be put into the pwg!"
[19:00] <jdahlin> Company, don't worry, I won't go away for the next couple of months :D
[19:01] <Company> i hope
[19:01] <Company> we need loads of new devs :)
[19:05] gheet ([email protected]) left irc: "Download Gaim [http://gaim.sourceforge.net/]"
[19:07] Action: taaz notes how nice it is to see people being paid to hack gstreamer.  only great things can happen like this. ;)
[19:08] Action: jdahlin nods
[19:08] <taaz> jdahlin: heh, of course you're happy about it ;)
[19:11] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[19:21] <ensonic> cu tomorrow, i'll started another run to build the plugins (crossing fingers)
[19:22] ensonic ([email protected]) left #gstreamer.
[19:22] pb_ ([email protected]) joined #gstreamer.
[19:26] sxpert ([email protected]) left irc: Read error: 110 (Connection timed out)
[19:35] Ura_gym ([email protected]) left irc: "Client exiting"
[19:37] wereHamster ([email protected]) joined #gstreamer.
[19:47] mathrick ([email protected]) joined #gstreamer.
[19:51] <mathrick> evening
[19:57] thomasvs ([email protected]) left irc: No route to host
[20:04] <ds-work> moo
[20:04] DrZaius ([email protected]) joined #gstreamer.
[20:04] Nick change: LeRoutier -> LeRaway
[20:06] somex1 ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[20:06] jdahlin ([email protected]) left irc: "Leaving"
[20:15] C ([email protected]) left irc: Remote closed the connection
[20:15] Uraeus ([email protected]) joined #gstreamer.
[20:17] <Uraeus> ello
[20:17] <mathrick> lo Uraeus
[20:18] trow ([email protected]) left #gstreamer ("Leaving").
[20:23] Thias ([email protected]) left irc: "yes, I'll get a life! :-)"
[20:34] BBB ([email protected]) joined #gstreamer.
[20:35] <mathrick> lo BBB
[20:36] <BBB> lo
[20:36] thomasvs ([email protected]) joined #gstreamer.
[20:37] <mathrick> yo thomasvs
[20:46] Nick change: LeRaway -> LeRoutier
[20:54] <BBB> thomasvs: any clue why my docs don't build?
[20:55] <thomasvs> BBB: paste your /etc/xml/catalog please to me privately
[20:55] <thomasvs> BBB: any idea why an svq3 wouldn't play with spider but plays when I do the pipeline manually ?
[20:55] <BBB> thomasvs: ehm... no... but I've seen it here, too... and it quite confuses me
[20:55] <thomasvs> me too :)
[20:56] <thomasvs> trying to fix up gst-thumbnail is giving me headaches
[20:56] <mathrick> thomasvs: could you help me with docs build too?
[20:56] <thomasvs> I don't think changing pipelines after running them works
[20:56] <thomasvs> yeah, but I need to eat first and then we have guests, so I don't know if I can do that today still, will tyr
[20:56] <mathrick> tyr?
[20:57] <BBB> docs used to build fine
[20:57] <BBB> but I'm now always getting errors for www.oasis-open.org failed to open
[20:57] <BBB> try
[20:57] Action: BBB reads through 600 emails
[20:57] <mathrick> BBB: same here, oasis-open.org
[20:58] <BBB> :/
[20:59] KoRnouille ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[20:59] taaz_ ([email protected]) joined #gstreamer.
[20:59] taaz ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[20:59] jimmy_dean ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[20:59] jimmy_dean ([email protected]) joined #gstreamer.
[20:59] KoRnouille ([email protected]) joined #gstreamer.
[21:01] sublett ([email protected]) left irc: Read error: 113 (No route to host)
[21:06] <mathrick> no way, can it really be?
[21:06] <mathrick> http://www.burtonini.com/blog//computers/docbook-xml-2004-02-02-15-03
[21:07] Action: mathrick updates packages list
[21:07] <bitshifter> hmmm, changing pipelines after running them is not supposed to work? So I'm not supposed to run filsrc ! typefind ! fakesink, set the state to NULL, replace typefind with oggdemux and vorbisdec, and start the pipeline again?
[21:07] <bitshifter> (as an example)
[21:08] <ds-work> it is supposed to work
[21:09] <bitshifter> supposed to, but doesn't really? or doesn't in some rare circumstances? :)
[21:09] trow ([email protected]) joined #gstreamer.
[21:10] apoc__ ([email protected]) joined #gstreamer.
[21:13] apoc_ ([email protected]) left irc: Read error: 110 (Connection timed out)
[21:16] <Company> bitshifter: does in some rare circumstances :)
[21:16] <thomasvs> BBB: and what exactly doesn't work ?
[21:17] <thomasvs> BBB: btw, any chance you could get mpeg to play back nicely again in the player before leaving ?
[21:18] <BBB> don't count on it
[21:18] <thomasvs> BBB: seriously ? oops
[21:18] <thomasvs> BBB: well, I'll fix your docs if you fix mpeg :)
[21:18] <BBB> I'm busy with packing and all that, and well.... just no single second
[21:18] <BBB> I can't even get myself to buy a new laptop right now
[21:18] <thomasvs> BBB: it'd be nice if you could braindump how ffmpeg stuff works so we can get someone to maintain it :)
[21:18] <BBB> sorry, can't
[21:18] <ds-work> BBB: do you have a monitor?
[21:19] <BBB> I'll braindump it when I arrive in NY
[21:19] <BBB> don't worry
[21:19] <BBB> I won't die
[21:19] <BBB> it'll just take a while
[21:19] <BBB> well, yes, that's how my laptop is being used right now
[21:19] <BBB> but I can't bring the monitor to the US
[21:19] <BBB> worst-case, I just bring my laptop and buy a cheap monitor somewhere over there
[21:21] <thomasvs> BBB: so I need a build log with your error to check
[21:22] <thomasvs> mathrick: what is the problem for you ?
[21:22] <BBB> it says 'oasis-open.org: host not found'
[21:22] <BBB> mathrick seems to have the same issue
[21:22] <mathrick> thomasvs: unable to contact oasis-open.org
[21:22] <thomasvs> BBB: dude, a log :)
[21:22] <thomasvs> BBB: not some error message from somewhere
[21:22] <thomasvs> I need to know when it says that and what it was doing
[21:22] <mathrick> thomasvs: but see URL i pasted minute ago, it may be finally fixed in Debian :)
[21:23] <BBB> ok, doing
[21:23] <thomasvs> mathrick: if you're on debian, you need to run jamesh's xml catalog rebuild script
[21:23] <BBB> gimme a minute, it first downloads the DTD
[21:23] <BBB> and that takes a while here
[21:24] <bitshifter> hmm, it would be great if someone could have a quick look at the valgrind output in http://bugzilla.gnome.org/show_bug.cgi?id=133181 and give a quick opinion whether that looks like a genuine gst problem, or whether that looks more like a 'you probably did things you're not supposed to do' problem
[21:24] <bitshifter> (I somehow don't really believe that something as simple as filesrc should be buggy)
[21:26] <BBB> jade:<URL>http://www.oasis-open.org/docbook/xml/4.2/dbpoolx.mod:3529:0:E: could not resolve host "www.oasis-open.org" (try again later)
[21:26] <BBB> jade:<URL>http://www.oasis-open.org/docbook/xml/4.2/dbpoolx.mod:3566:3:E: parameter entity "tbl.table.att" not defined
[21:26] <BBB> and then a load of errors about all sort of undefined tags (the one you get if you build offline)
[21:27] <ds-work> filesrc isn't that simple
[21:29] <BBB> it's not filesc
[21:29] <BBB> it's the videostream typefind implementation
[21:29] <BBB> I wrote that
[21:29] <BBB> there's a good chance there's a bug in it
[21:30] <thomasvs> BBB: seriously, I need *a log*
[21:30] <thomasvs> BBB: I need to know *what dir*, *what command*, *what it's doing*
[21:30] <BBB> what's a log?
[21:30] <thomasvs> BBB: not just a random paste
[21:30] <BBB> docs/pwg
[21:30] <thomasvs> output of the commands
[21:30] <BBB> make
[21:30] <BBB> and that's the error
[21:30] <thomasvs> BBB: cd docs/pwg; make > log 2>&1
[21:31] <thomasvs> send me that
[21:31] <bitshifter> k. Thanks for the comments.
[21:37] <thomasvs> hm, pngenc doesn't seem to work well anymore
[21:37] Zeenix ([email protected]) joined #gstreamer.
[21:37] <ds-work> Company: should gst_uri_protocol_check_internal() be using g_ascii_isalpha() instead of isalpha()?
[21:37] <Zeenix> hello
[21:38] <Company> ds-work: yeah
[21:38] <Company> ds-work: it should work locale intependently at least
[21:39] <ds-work> just to check that I'm thinking correctly, we should _never_ be using ctype.h stuff, since it doesn't work with ASCII (when in a non-C locale) or UTF-8 (since it doesn't handle wide chars)
[21:40] bitshifter ([email protected]) left irc: "(null)"
[21:40] bitshifter ([email protected]) joined #gstreamer.
[21:44] <thomasvs> gst-launch-0.7 videotestsrc ! pngenc ! filesink location=test.png doesn't terminate
[21:44] <thomasvs> though pngenc sends EOS
[21:44] <thomasvs> any idea what could be wrong ?
[21:45] <ds-work> videotestsrc doesn't get EOS
[21:45] <thomasvs> ok, so the pipeline doesn't stop until all elements get eos
[21:46] <thomasvs> so what do I fix in videotestsrc ? just "you asked for eos, I'm setting myself to eos" ?
[21:46] <ds-work> videotestsrc is doing what it is supposed to do
[21:46] <ds-work> it can't be expressed with gst-launch
[21:46] <ds-work> except with ctrl-c
[21:47] <ds-work> or -i100
[21:47] <thomasvs> ok, I don't follow.  I expect the pipeline above to stop after pngenc signals EOS, which it will after a complete frame.
[21:47] <thomasvs> what's wrong with my expectation ?
[21:47] <ds-work> pngenc continues to listen to videotestsrc after that
[21:48] <ds-work> waiting for you to signal it to grab another frame
[21:48] <ds-work> iirc
[21:48] <thomasvs> so, if I just want one frame, what should I do programatically ?
[21:48] <Company> eEeEeEk
[21:49] <ds-work> thomasvs: -i10, but that's a hack
[21:49] <Company> question 1: why does filesrc not use mmap?
[21:49] <Company> question 2: why does filesrc not set timestamps etc when not using mmap?
[21:49] <Company> (i'll fix 2, but someone explain 1 to me)
[21:49] <thomasvs> BBB: put it online somewhere, dcc don't work here
[21:49] <thomasvs> ds-work: uhm, there must be some programming way to do this ?
[21:50] <ds-work> Company: is your question "why isn't filesrc using mmap on this file"?
[21:50] <Company> ds-work: yes
[21:50] <ds-work> Company: where's the file?
[21:50] <Company> ds-work: VFAT partition
[21:50] <thomasvs> mmap don't work off vfat iirc
[21:50] <ds-work> probably doesn't support mmap
[21:50] <Company> oh?
[21:51] <ds-work> thomasvs: it's not possible with gst-launch
[21:51] Action: mathrick didn't know mmap() was partition type dependant
[21:51] <thomasvs> ds-work: hm, ok.  so a short program instead then, and connect to eos on filesink ?
[21:52] thaytan ([email protected]) left irc: Read error: 60 (Operation timed out)
[21:52] Nick change: apoc__ -> apoc
[21:52] <apoc> hi
[21:52] <mathrick> yo apoc
[21:53] <ds-work> thomasvs: yes
[21:54] <Company> nope, vfat must do mmap
[21:57] <mathrick> Company: hmm, it may be true, bittorrent's mem usage suddenly dropped by ~50MB per instance when I exhausted space on my VFAT partitions and started to download to ext3 ones
[22:00] robUx4 ([email protected]) joined #gstreamer.
[22:00] <robUx4> lo
[22:00] <mathrick> lo robUx4
[22:01] <Company> filesrc only does non-mmaped access since July
[22:01] <Company> i have always accessed my music on a vfat partition via filesrc
[22:01] <Company> always = last 3 years
[22:03] dolphy ([email protected]) joined #gstreamer.
[22:05] <BBB> thomasvs: can I email it?
[22:06] <thomasvs> BBB: sure
[22:06] <thomasvs> hi robUx4
[22:06] <thomasvs> robUx4: do you have a patch ?
[22:06] <BBB> done
[22:06] <Company> yep, vfat uses mmap
[22:06] <Company> why could mmap fail?
[22:06] <thomasvs> lots of reasons, check return value and errno
[22:07] <robUx4> thomasvs: almost yes
[22:07] <robUx4> I just figured out the job was not complete :/
[22:07] <Company> yeah, uhm, funny
[22:07] <Company> that happens on the 500th file :o
[22:08] <ds-work> perhaps a buffer isn't getting freed, and you're running out of address space
[22:09] Action: ds-work stabs in the dark
[22:11] Nick change: KA -> KA_afk
[22:13] <thomasvs> Company: just write an app that opens them one by one, seeks through them completely, then goes on to the next ?
[22:13] <thomasvs> should get there fast enough
[22:13] <Company> yeah, it looks like a mem issue
[22:13] <Company>  2676 root      18   0  589M 290M  555M D    45.7 57.7   0:21 detect
[22:13] <Company>  2676 root      18   0  738M 325M  695M D    50.1 64.6   0:28 detect
[22:13] <dolphy> BBB: i m almost done optimizing X Events in x[v]imagesink
[22:14] <thomasvs> Company: does osssink implement correct clocking behaviour now ?
[22:14] <dolphy> BBB: i m taking all the configure events out of the queue and only take the last one
[22:15] <Company> thomasvs: dunno, it should, but i haven't tested for a while - i still have a nice green image in cvimagesink :/
[22:15] <thomasvs> Company: ok, we should get that fixed :) does ffcolorspace ! ximagesink work better ?
[22:17] <Company> nope, ximagesink doesn't work either
[22:17] <thomasvs> Company: what caps does it settle on ?
[22:19] <Company> i haven't debugged it
[22:19] <BBB> dolphy: sounds good... make sure that you don't forget events ;)
[22:19] Nick change: Jara[zZ] -> Jaramir
[22:19] <BBB> i.e., flush all resize events
[22:19] <BBB> but not others
[22:19] <Company> i just thought it's a common issue
[22:19] <BBB> etc.
[22:20] kmaraas ([email protected]) joined #gstreamer.
[22:20] <dolphy> BBB: yeah
[22:20] <dolphy> BBB: but that's not optimizing that much
[22:20] <dolphy> BBB: because the handle_xevents function is run quite often
[22:20] <dolphy> BBB: 25fps
[22:20] <dolphy> BBB: so events are not really flooding the queue :)
[22:21] <thomasvs> BBB: cat /usr/share/sgml/docbook/xmlcatalog
[22:21] <thomasvs>  to me privately please
[22:21] <thomasvs> BBB: as well as rpm -qa | grep docbook
[22:21] <ds-work> have we ever fixed the "x[v]imagesink doesn't update when paused"?
[22:21] <ds-work> or with videotestsrc framerate=1.0?
[22:22] <dolphy> ds-work: doesn't update ?
[22:22] <dolphy> ds-work: what do you mean ?
[22:22] <Company> no we haven't
[22:22] <ds-work> gst-launch-0.7 videotestsrc ! video/x-raw-rgb, framerate=1.0 ! ximagesink
[22:22] <ds-work> move another window on top of the video window
[22:23] <ds-work> it takes ~1 sec to redraw
[22:23] <ds-work> or, with xvimagesink, move the window
[22:23] <dolphy> well yeah
[22:23] <dolphy> that's quite normal
[22:25] <ds-work> normal, but wrong
[22:25] <dolphy> i guess every X application as to draw on expose
[22:25] <dolphy> x[v]imagesink can't draw on expose
[22:25] thaytoo ([email protected]) joined #gstreamer.
[22:25] <ds-work> this is going to be a problem with swfdec
[22:26] <ds-work> since it may want navigation events for a still frame
[22:26] <ds-work> s/may want/will want/
[22:26] <thomasvs> dolphy: why can't it draw on expose ?
[22:26] <Company> because it's never scheduled
[22:26] <dolphy> because it would mean having a separate thread polling events
[22:27] <dolphy> catching expose event
[22:27] <dolphy> and putting current image to the drawable
[22:27] <dolphy> that's very easy to do
[22:27] <dolphy> but not gstreamerish at all
[22:27] <Company> that's why i've always said events should be sent by the application
[22:28] <Company> an element can't do anything that is not pipeline specific
[22:28] yippi ([email protected]) joined #gstreamer.
[22:28] <Company> because it will only run when the scheduler decides it should
[22:28] <dolphy> well and then
[22:29] <dolphy> you would do something like gst_x_overlay_send_expose_event (GST_X_OVERLAY(video_sink)); ?
[22:29] <Company> no, i would do something like gst_overlay_move or _resize
[22:30] <dolphy> that doesn't fix the problem
[22:30] <Company> or maybe just gst_overlay_redraw
[22:33] <dolphy> that's one possibility
[22:33] <dolphy> not so sexy though
[22:33] <dolphy> Company: have you looked into the timing issue ?
[22:34] <thaytoo> morning, all
[22:34] <thaytoo> sounds like I arrived on time
[22:35] <Company> telling an element to do anything but running inside its pipeline is a lot less sexy to me
[22:35] <Company> dolphy: no, i just have a green 16x16 window in gst-player
[22:36] <dolphy> Company: do you have visualization turned on ?
[22:36] <BBB> thomasvs: see email
[22:36] <BBB> oh man I hate email
[22:37] sublett ([email protected]) joined #gstreamer.
[22:37] <Company> dolphy: nope
[22:41] <dolphy> Company: using ximagesink ?
[22:42] yippi ([email protected]) left irc: Remote closed the connection
[22:43] <BBB> does anyone mind if I commit some documentation that's not checked?
[22:43] <BBB> it succeeds the XML validation
[22:43] <BBB> but I can't compile docs
[22:43] <BBB> I think it works
[22:44] <Company> dolphy: both ximagesink and xvimagesink don't work
[22:45] <dolphy> Company: can't reproduce
[22:45] <dolphy> Company: works just fine here on both gst-player and totem
[22:46] <dolphy> Company: does examples/gstplay/player works ?
[22:46] <BBB> Company: do you get 16x16 windows?
[22:46] <BBB> Company: that's what I get
[22:47] <thomasvs> BBB: what platform ? rh9 ?
[22:47] <dolphy> eveybody gets a 16x16 window
[22:47] <Company> BBB: yeah
[22:47] <dolphy> but then it resizes to the video size
[22:47] <Company> dolphy: do you use most recent core and plugins?
[22:47] <dolphy> that's the fixate issue
[22:47] <dolphy> Company: they are 20 minutes old
[22:47] <ds-work> gstmediaplay.h:29:27: gst/play/play.h: No such file or directory
[22:48] <thomasvs> ds-work: did you update and install gst-plugins ?
[22:48] <BBB> thomasvs: yes
[22:48] <ds-work> thomasvs: duh, doing that now
[22:49] <BBB> dolphy: it doesn't resize here
[22:49] <BBB> and fixation should never happen... that's a plugin bug... it should delay
[22:49] <BBB> (imo)
[22:49] <BBB> but anyway
[22:49] <BBB> it doesn't resize
[22:50] <dolphy> BBB: gst-plugins/examples/gstplay/player matrix.avi does not resize ?
[22:51] <BBB> no
[22:51] <BBB> it doesn't resize for whichever movie I throw at it
[22:51] <dolphy> huhu
[22:51] <ds-work> thomasvs: could you comment on http://bugzilla.gnome.org/show_bug.cgi?id=133086
[22:51] <dolphy> any other people with the same issue ?
[22:52] Action: BBB will bbl
[22:52] <BBB> sleep time for me
[22:52] <ds-work> gst-player works here, except for mpeg
[22:52] <BBB> need to continue this hell of moving-to-us tomorrow
[22:52] <dolphy> yeah mpeg is broken everywhere
[22:52] <BBB> (with a broken laptop, grmbl)
[22:52] <BBB> bye
[22:52] BBB ([email protected]) left irc: "Client exiting"
[22:54] <thomasvs> ds-work: he's probably correct, I'll remove it
[22:54] <dolphy> ds-work: if gst-player works for you please load 3 or more videos in the playlist
[22:55] <dolphy> ds-work: play the first one a little
[22:55] <dolphy> ds-work: then press next
[22:55] <dolphy> ds-work: and you will see how gstreamer timing is broken atm
[22:59] Action: ds-work notices that swfdec is broken
[22:59] <ds-work> maybe not
[23:00] <ds-work> GStreamer-CRITICAL **: file gstpad.c: line 2867 (gst_pad_push): assertion `data != NULL' failed
[23:00] <ds-work> #8  0x41089a1f in gst_switch_loop (element=0x83b16c0) at gstswitch.c:209
[23:00] <ChrisHJW> !seen ploum
[23:01] <dolphy> ds-work: that was with an audio only file ?
[23:01] <ds-work> is GStreamer officially i18n-ified yet?  #114216
[23:01] <ds-work> dolphy: no, flash
[23:01] Nick change: harshyWork -> harshy
[23:03] <thomasvs> ds-work: yep, and it works
[23:03] <thomasvs> ds-work: want screenshots ? :)
[23:03] <thomasvs> http://thomas.apestaart.org/download/screenshots/gst-player-error-2.png
[23:05] <ds-work> then I'm closing #114216
[23:05] <dolphy> thomasvs: so what should we do about volume ?
[23:06] <dolphy> thomasvs: put 2 of them ?
[23:07] <thomasvs> dolphy: I think you should yeah
[23:07] <thomasvs> dolphy: the actual volume controlling output volume should be after the queue, in any case
[23:07] <robUx4> extern GstElement *_gst_parse_launch (const gchar *, GError **);
[23:07] <robUx4> where is the actual code ?
[23:07] <thomasvs> dolphy: and the vis volume should also be after a queue
[23:07] <Company> robUx4: gst/parse/grammar.y
[23:07] <thomasvs> robUx4: generated by parse.y iirc
[23:08] <thomasvs> ehm, yeah, that one :)
[23:08] <robUx4> doh !!!!
[23:08] Action: robUx4 cries
[23:08] <Company> robUx4 needs to install bison and flex on windows now ;)
[23:08] <robUx4> whenever I think I'm close to the end, there are other troubles :/
[23:08] <robUx4> lol
[23:08] <robUx4> yeah
[23:08] <robUx4> you think I'm going to ?
[23:09] <Uraeus> robUx4: we are sure you will :)
[23:09] <Company> sure :p
[23:09] <robUx4> nah
[23:09] taaz ([email protected]) joined #gstreamer.
[23:09] <robUx4> why does it need to be processed ?
[23:10] <robUx4> somebody needed to try bison and flex for his studies ?
[23:10] taaz ([email protected]) left irc: Remote closed the connection
[23:10] taaz_ ([email protected]) left irc: "Disconnecting"
[23:11] <mathrick> robUx4: don't windoze people use parser generators?
[23:13] Action: ChrisHJW worries about the matroska lead developer ... porting this beats seems to be troublesome ...
[23:13] <robUx4> when needed of course
[23:13] <ChrisHJW> s/beats/beast
[23:13] <robUx4> now I have this nice function returning NULL
[23:13] eventide ([email protected]) joined #gstreamer.
[23:13] <robUx4> and everything compiles :)
[23:13] <robUx4> that wasn't that hard ;)
[23:14] <Zeenix> should'nt an xoverlay implementing element keep a record of whether it's running in pasive or active mode so that i can destroy it's window in the end..?
[23:15] <dolphy> thomasvs: ok i m moving volume in the audio thread for the moment
[23:15] <dolphy> thomasvs: anyway visualization is still not working
[23:15] <dolphy> thomasvs: we'll see later
[23:15] <mathrick> robUx4: I may be wrong, but it's common to use parser generators to generate parsers, isn't it? ;) And *_gst_parse_launch() looks awfully lot like parser :P
[23:18] <robUx4> well, I never used such a thing
[23:18] <robUx4> now gst-launch "runs"
[23:18] <dolphy> thomasvs: ok i fixed the open menu in the player
[23:18] <dolphy> thomasvs: fixed events in x[v]imagesink
[23:19] apoc_ ([email protected]) joined #gstreamer.
[23:19] <dolphy> thomasvs: moved volume in audio thread and brought audioconvert back in the gstplay pipeline
[23:20] <ChrisHJW> <robUx4> now gst-launch "runs"  <--- where to :P ?
[23:20] <thomasvs> robUx4: coolness
[23:21] <thomasvs> robUx4: try gst-launch fakesrc ! fakesink silent=false
[23:21] <robUx4> I just don't know a good command-line...
[23:21] <robUx4> ah, ok
[23:21] <mathrick> robUx4: great :)
[23:21] <thomasvs> robUx4: see if it prints out stuff
[23:21] <wereHamster> What's the reason to implement your own scheduler? With the new O(1) scheduler in the 2.6 kernel there is no need to du it (kinda reinventing the wheel)
[23:21] <thomasvs> wereHamster: different kind of schedulers
[23:21] <robUx4> "ERROR: pipeline could not be constructed"
[23:22] <thomasvs> robUx4: run gst-register first
[23:22] Action: thomasvs is pretty sure that will blow up
[23:22] <ds-work> argh!  Mandrake is creating a 3rd Debian
[23:22] <dolphy> time to sleep
[23:22] <dolphy> night 
[23:22] <thomasvs> ds-work: they are ? where ?
[23:22] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[23:23] <ds-work> maybe not
[23:23] <mathrick> ds-work: 3rd?
[23:23] <ds-work> mathrick: fedora
[23:23] <mathrick> ds-work: is it so Debianish (it == fedora) ?
[23:24] <ds-work> "community-created distribution", yes
[23:25] <robUx4> thomasvs: I just want to finish for today
[23:25] <robUx4> so I'm going to publish the sources I have (modified)
[23:25] <robUx4> and explain how to use it
[23:25] DrZaius ([email protected]) left irc: "Client exiting"
[23:25] apoc ([email protected]) left irc: Read error: 60 (Operation timed out)
[23:25] <robUx4> I also have to publish my GLib modified
[23:25] <ChrisHJW> robUx4 : you have an account on freedesktop.org ?
[23:27] <ds-work> robUx4: attach the patch to a bug, and I'll start merging
[23:27] <robUx4> ChrisHJW: no
[23:27] <robUx4> I'm just going to do a zip for both
[23:27] <robUx4> and send an email with some explanations
[23:28] <ds-work> diff -u is 100x better
[23:28] <LeRoutier> -uN
[23:29] <robUx4> on UNIX probably
[23:29] <robUx4> and you'll be happy to be able to do it with the sources I'll publish :D
[23:29] <Company> ds-work: currently buffers created by filesrc ref the filesrc
[23:30] <Company> ds-work: any objection to me changing that and use the DONTKEEP stuff?
[23:30] Nick change: apoc_ -> apoc
[23:30] <ds-work> ds-work: that seems somewhat reasonable
[23:30] <ds-work> heh
[23:30] <ds-work> Company: what is wrong with reffing filesrc?
[23:30] water ([email protected]) joined #gstreamer.
[23:31] water ([email protected]) left #gstreamer ("The water has parted.").
[23:31] <ds-work> robUx4: don't you have cvs installed?
[23:31] <Company> ds-work: http://bugzilla.gnome.org/show_bug.cgi?id=133181 is the perfect example
[23:31] <robUx4> what do you mean CVS installed ?
[23:31] <robUx4> a client ?
[23:32] <ds-work> robUx4: yes
[23:32] taaz ([email protected]) joined #gstreamer.
[23:32] <Company> ds-work: it nicely refs buffers from every filesrc and doesn't free them while freeing the filesrc
[23:32] <Company> s/freeing the filesrc/unreffing the filesrc/
[23:33] <Company> ds-work: so in the end there's huge mmaps and filesrcs lying around
[23:33] <robUx4> ds-work: sure, I do
[23:33] <ds-work> Company: that sounds like a bug, not a theoretical problem
[23:34] <ds-work> robUx4: then you can do 'cvs diff -u'
[23:34] <robUx4> [email protected] <- is that the correct list ?
[23:34] <ds-work> yes
[23:34] <Company> ds-work: does closing the file munmap the data?
[23:34] <ds-work> Company: no
[23:34] <robUx4> ds-work: Windows user are lazy ! I don't use CVS with a command line anymore. Not even sure I can
[23:35] <Company> in that case we don't need to ref the filesrc
[23:35] <ds-work> Company: hrm, true.  do that :)
[23:36] <mathrick> robUx4: point is, most windoze CVS clients are really GUIs for commandline cvs
[23:37] <LeRoutier> good night ppl
[23:37] <mathrick> nite LeRoutier
[23:37] LeRoutier ([email protected]) left irc: "Leaving"
[23:37] <Company> ds-work: can you mmap the same region twice?
[23:38] <robUx4> I know
[23:38] <ds-work> Company: yes
[23:38] <robUx4> I just don't want to spend 1/2h firguring out (now) how to make it work
[23:38] <ds-work> instead, we have to spend 1/2 hour to extract a useful patch from your zip file?  thanks
[23:40] Action: taaz still thinks we should be using -p diff option all the time ;)
[23:41] <mathrick> robUx4: it's easy. google for cvs, install, run cvs diff -u. I installed cvs on win2k3 recently, was really straightforward task (running unzip in fact is enough)
[23:41] <mathrick> taaz: what does it mean?
[23:41] <taaz>        -p  --show-c-function
[23:41] <taaz>               Show which C function each change is in.
[23:42] <robUx4> it's just that I'm already doing something else (preparing to publish my changes)
[23:42] <mathrick> heh
[23:42] <robUx4> the CVS will come later, maybe tomorrow
[23:42] <robUx4> after someone explain me what needs to be done
[23:42] <taaz> it's not always right, but still useful
[23:42] <robUx4> and as I made many changes I want some of you have a look first before applying to CVS
[23:43] <ds-work> robUx4: I'm about 10x more likely to look at a patch than a zip file
[23:43] <ds-work> in fact, I _won't_ look at a zip file
[23:44] <ds-work> I'll just wait for someone else to create a patch
[23:44] <robUx4> your decision
[23:44] <robUx4> email sent now
[23:44] <ds-work> that's just how things work around here
[23:44] <robUx4> I know
[23:45] <robUx4> right now my priority is to have other ppl on Windows get their hands on the code and make it work further
[23:48] trow ([email protected]) left irc: "Leaving"
[23:51] <wereHamster> good night...
[23:51] wereHamster ([email protected]) left #gstreamer.
[23:53] <robUx4> now were is my email gone :(
[23:55] <ds-work> sf is slow
[23:56] <robUx4> maybe my provider too
[23:56] <robUx4> I sent it to matroska-devel too which is supposed to be fast
[23:57] <ds-work> did you attach the zip files?  most mailing lists moderate anything over 40 kB
[23:57] <robUx4> nop
[23:57] <ds-work> just showed up here
[23:57] <robUx4> :)
[00:00] --- Tue Feb  3 2004
[00:00] <Uraeus> night :)
[00:00] Uraeus ([email protected]) left irc: "Client exiting"
[00:02] <apoc> night ppl :)
[00:02] apoc ([email protected]) left irc: "Leaving"
[00:03] <robUx4> nite too
[00:03] robUx4 ([email protected]) left irc: "Baby wants to bleep"
[00:05] <Zeenix> i am getting this error from gcc: libgstreamer-0.7.la: file not recognized: File format not recognized
[00:06] <ds-work> Zeenix: do a fresh autogen and recompile
[00:06] <Zeenix> ok
[00:07] <Zeenix> ds-work: but i am compiling against unistalled gst
[00:08] <ds-work> Zeenix: it's a libtool problem
[00:09] ChrisHJW ([email protected]) left irc: Connection timed out
[00:10] <Zeenix> ds-work: you mean a fresh update too?
[00:10] <ds-work> no
[00:15] Shoragan ([email protected]) left irc: "Leaving"
[00:36] markey ([email protected]) left irc: "nite"
[00:37] herzi ([email protected]) left irc: Read error: 54 (Connection reset by peer)
[00:37] Action: Company just deleted a bunch of lines from filesrc, including a mutex
[00:38] herzi ([email protected]) joined #gstreamer.
[00:39] mathrick ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[00:40] mathrick ([email protected]) joined #gstreamer.
[00:53] <Company> ds-work: shouldn't gst_caps_debug be removed/deprecated?
[00:57] <Company> am i missing something in http://bugzilla.gnome.org/show_bug.cgi?id=133065 ?
[00:57] <Company> there's no soundcard.h include anywhere
[00:58] <ds-work> Company: gst_caps_debug() is a stupid #define in gstpad.c
[00:58] <ds-work> perhaps GST_DEBUG_CAPS?
[00:58] <Company> that too
[00:58] <Company> i got a message from the docs build about it not having docs
[00:58] <Company> didn't look further
[01:00] <ds-work> Company: configure.ac looks for it
[01:01] <ds-work> er, nm
[01:03] jcsston ([email protected]) joined #gstreamer.
[01:04] <ds-work> Company: it was there in 0.7.3
[01:04] <Company> so i'll just close the bug now :)
[01:04] <ds-work> just did
[01:07] <Company> ah, i can't read cvs log output
[01:08] <ds-work> because it's not there
[01:08] <ds-work> or not mentioned specifically
[01:08] <Company> no i thought all the tags listed applied to the most recent version...
[01:09] <ds-work> in 'cvs log'?  the tags all show the version they point to
[01:10] <Company> yeah, i said i can't read :p
[01:15] <bitshifter> Company: hmm, so I am supposed to do what with the filesrc? unref it once for every buffer the filesrc creates?
[01:17] <Company> bitshifter: you keep the buffer around and don't unref it in your tagging element
[01:17] <bitshifter> oops.
[01:17] <bitshifter> hmm
[01:17] <Zeenix> ds-work: i am still getting the same error (after a fresh autogen & re-compile)
[01:20] <Company> bitshifter: hm, or do you unref the buffer after filesrc is already gone?
[01:20] <Company> bitshifter: (btw, i haven't committed my fix yet)
[01:20] Action: Company hides
[01:21] <Company> i changed another thing i need to check first by compiling gst-plugins...
[01:21] <bitshifter> Company: hmm, no, you are right, I think I don't unref the tagfilter->backlog buffer when the element is destroyed.
[01:24] pb_ ([email protected]) left irc: "later"
[01:24] <bitshifter> I guess not noticing a 1.2GB leak is a sure sign that I bought a bit too much RAM :D
[01:25] <Company> no, that's not a problem
[01:25] <Company> linux doesn't map all that memory
[01:25] <bitshifter> no?
[01:25] <Company> it just shows up in top :)
[01:26] <Company> run top in one window and detect in another one
[01:26] <Company> .. and enjoy :)
[01:26] <ds-work> for the hell of it, I mapped the entirety of a 40 GB drive once on my alpha
[01:26] <bitshifter> thanks 
[01:26] <mathrick> Company: what is detect?
[01:27] <Company> mathrick: bitshifters tool attached to his bugreports
[01:27] <Company> report
[01:27] <mathrick> Company: ic
[01:29] <Company> but it doesn't get over 16MB data mmaped now
[01:31] <Company> Elapsed: 477.3 seconds for 563 files = 0.8 seconds/file
[01:31] <ds-work> erm
[01:31] <ds-work> that's not particularly speedy
[01:31] <Company> that's not bad considering i had to load every file from disk and it printed 5 linesa of output for every file :)
[01:35] <mathrick> Company: how's it when you do > /dev/null ?
[01:37] <mathrick> Company: modulo you have files in cache now, so it isn't very reliable way of measuring
[01:44] <Company> /dev/null doesn't work, the output is sent there, too ;)
[01:45] <Company> ds-work: are GPL autoconf macros ok?
[01:46] <ds-work> Company: most autoconf stuff is GPL
[01:47] <mathrick> Company: hmm, that's what I intended to, send output to /dev/null :). ie., remove I/O overhead from running time
[01:47] <ds-work> with the notice that using autoconf with a non-GPL project is considered aggregation, not derivation
[01:47] <Company> ds-work: i mean downloading them from the web to use in gst autoconf
[01:47] <ds-work> autoconf itself is GPL
[01:48] <ds-work> so GPL macros are never a problem
[01:48] <ds-work> iirc
[01:49] <ds-work> you might want to verify that with thomasvs
[01:50] <Company> thomasvs: are GPL autoconf macros ok in gstreamer?
[01:53] <bitshifter> mathrick: for what it's worth: I get 0.2 secs/file for 1071 files (ext3fs); and 0.14 secs/file with output to >/dev/null
[01:53] <bitshifter> not that that actually means anything
[01:54] <bitshifter> :)
[01:54] <Company> bitshifter: try gst-typefind-0.7 /path/to/dir/*.mp3
[01:55] <mathrick> bitshifter: but anyway you're 1 minute ahead, aren't you ;)
[01:55] <Company> that's 17.2 secs (user) for 563 files
[01:55] <Company> time gst-typefind obviously :)
[01:56] <bitshifter> Company: gst-typefind takes 50 secs her (real) and 21 secs (user). But then, gst-typefind cheats in a way ;)
[01:56] <mathrick> wow, gst-typefind is speedy
[01:56] <bitshifter> * here
[01:57] <mathrick> bitshifter: howso?
[01:57] <mathrick> bitshifter: ie, what cheats?
[01:58] <bitshifter> well, if there's an ID3v2 tag right at the beginning fo the file, gst-typefind just says the mime type is application/x-id3, while my program would then go and put in an ID3 tag filter and run again to find the actual content-type (e.g. audio/mpeg)
[01:58] <bitshifter> so it's not really comparable
[01:58] <Company> yeah
[01:59] <Company> the interesting thing about gst-typefind is that it's the only element we have that doesn't use a new pipeline for the next file
[01:59] <Company> so it's really more like "what speed can i expect from a working gstreamer?" :)
[01:59] <bitshifter> nice
[02:00] <bitshifter> is it that expensive to set up a pipeline?
[02:01] <Company> no idea how expensive it really is
[02:01] <Company> it shouldn't be that much
[02:03] <Company> but it could be noticable if you do such things
[02:03] Action: Company tests
[02:03] <bitshifter> heh, once I put the hashing in it won't make much difference any more ;)
[02:04] <Company> 17.8 secs
[02:04] <Company> you lose 1 msec per file
[02:05] <mathrick> not that much of deal
[02:05] <bitshifter> nah
[02:08] <Company> well, it's 3%
[02:09] <ds-work> user    0m0.560s
[02:09] <ds-work> creating/destroying 100 fakesrc ! fakesink pipelines
[02:10] <Company> it only gets interesting with bigger pipelines...
[02:10] herzi ([email protected]) left irc: Remote closed the connection
[02:10] <ds-work> of course
[02:12] <ds-work> hopefully 2.5 ms per element is reasonable, though
[02:14] herzi ([email protected]) joined #gstreamer.
[02:23] pippin ([email protected]) joined #gstreamer.
[02:24] <Zeenix> ds-work: so what else you think could i try?
[02:26] <pippin> is there any figures on hwo many gst people will attend guadec this year?
[02:34] Nick change: Jaramir -> Jara[zZ]
[02:37] sublett ([email protected]) left irc: Remote closed the connection
[02:43] <Company> pippin: no idea, Uraeus recently posted to the list, but there wasn't much response
[02:44] <Company> pippin: though I bet Uraeus, thomasvs, dolphy and me will be there
[02:44] <Company> dunno about ds and BBB
[02:45] <pippin> I'm pondering to do a presentation on a project I've been tinkering with,. and I'd like to step on as few toes as possible :)
[02:46] ChrisHJW ([email protected]) joined #gstreamer.
[02:47] <Company> what project?
[02:48] <pippin> Company: I'm hacking on a compositing engine,. which I plan to base on GEGL (the gimp compositing engine to be) http://schweden.mine.nu/bauxite/screenshots/bauxite_0_71.png
[02:50] <Company> go ahead
[02:50] <Company> we'll just wrap it in the end if it's useful :p
[02:50] <pippin> :)
[02:51] <Company> i personally don't have any problems with people doing similar stuff to gstreamer and i don't think others have either
[02:51] <Company> though we obviously encourage code reuse - we practice that ourselves nicely
[02:51] <pippin> Company: the way I see it,. the goals of gegl and gstreamer are overlapping, but different,. and ultimately,. gstreamer should have gegl nodes for doing compositing work
[02:52] <Company> yeah, it'll probably end that way
[02:52] <Company> gegl is not about doing realtime image compositing, is it?
[02:52] <pippin> IMO it should be
[02:53] <Company> that could indeed become interesting
[02:54] <pippin> I'm doing realtime compositing at the moment,. but then again,.. gegl doesn't really exist yet and I'm cheating,. but there is not reason gegl shouldn't be as efficient as what I've got at the moment
[02:55] <Company> well, from looking at gegl it looks a lot like gstreamer with video only
[02:56] <pippin> looking at gegl, is not looking at my screenshots
[02:56] <Company> gegl.org
[02:56] <pippin> yep,.
[02:56] <Company> browsing around there
[02:56] <pippin> not very up to date,.
[02:56] <Company> gstreamer has the problem of being _very_ generic
[02:57] <Company> which is obviously its greatest strength, too
[02:57] <pippin> gegl will probably be better suited at optimization of compositing,. cache structures,.. tile management,. color space and depth issues etc
[02:58] <Company> well, you could do all of that inside gst, too
[02:59] <Company> gstreamer itself (not the plugins) just decides if the pipeline connections fit and which element to schedule next


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn