IRC Logs

*******************************************************************
[03:15] walters ([email protected]) left irc: "out"
[03:16] <ds-work> ah
[03:16] <ds-work> btw, why do they have request pads?
[03:23] <Company> because fakesrc and fakesink were used as testbeds for every new technology gstreamer got in the wtay and omega time :)
[03:26] <Company> it's the videotestsrc of last year ;)
[03:26] <ds-work> I'd like to rip them out of core, put them in gst-plugins/gst/debug, and call them debugsink and debugsrc
[03:26] <ds-work> and replace with simple fakesrc and fakesink
[03:27] harshy ([email protected]) left irc: "I quit for now"
[03:28] <Company> go ahead
[03:29] <Company> im want an element that randomly changes ~2% of the bytes in a stream
[03:29] <Company> let it save the seed somewhere so it's reproducable and run it with the same file for a night
[03:29] <ds-work> iirc, identity can do that :)
[03:29] <Company> if it never crashes your element wins :)
[03:30] <Company> the only thing we need for that are small test streams
[03:35] Action: ds-work just now realizes that gst_element_error() adds a prefix to the 2nd and 3rd arguments
[03:51] <thaytoo> Is my DVD stuff so crap that _noone_ is prepared to reply to it? :-/
[03:52] <ds-work> thaytoo: does it work?
[03:52] <ds-work> thaytoo: is it coded well?
[03:53] <thaytoo> ds-work: it works so far
[03:53] <thaytoo> as for coded well... I'm not sure
[03:53] <thaytoo> and I want advice on how to extend it
[03:53] <thaytoo> for example, how to set things up so that selecting different audio/video/subtitle tracks in a DVD menu can be implemented
[03:54] <Company> that's a tough thing we need to sort out
[03:54] <Company> because that has to work for subtitle selection in these weird subtitling formats (ogm etc), too
[03:55] <jimmy_dean> ds-work: the author of the LADSPA plugins emailed me back...he thinks I should just use his plugin so that when he updates and improves them, then there aren't any offshoots that would need separate updating
[03:55] <thaytoo> mathrick's suggestion of roping switch in appeals, but I'm not sure
[03:55] <Company> and i'll look at it but i'm probably not very qualified to comment since i don't even on a dvd drive...
[03:56] <Company> and i don't own one either
[03:56] <Company> i dunno about switch
[03:56] <Company> switch doesn't allow you to get information about the track
[03:56] <Company> so siwtch can just say "select one from this list: 1, 2, 3"
[03:57] <thaytoo> yeah, only mpegdemux really knows that information at the moment
[03:58] <Company> you could put it inside a tag
[03:58] <Company> put that gets ugly real fast when you try to autoplug this
[03:59] <Company> (esp. when going back to the general format case)
[03:59] <jimmy_dean> ds-work: for the equalizer plugin of course...forgot to mention that
[03:59] <sri> I'm getting this with gstreamer head, when I run gst-mixer: gst-mixer: relocation error: /gnome2/lib/gstreamer-0.7/libgstadder.so: undefined symbol: gst_static_pad_template_get
[03:59] <sri> anybody know what that means?
[04:00] <sri> this of course kinda breaks my gnome-volume-* stuff.. :/
[04:04] <ds-work> sri: it means you have a >1-month-old version of gstreamer
[04:05] <ds-work> Company: we need a switchbin -- which switches between multiple pipelines instead of switching between sink pads
[04:06] <sri> ds-work: strange..I've been getting the code from jhbuild, which I think is getting it from freedesktop.org
[04:07] <ds-work> ... ! switchbin( bin(warpTV) bin(videobalance) bin(some_other_filter)) ! xvimagesink
[04:07] <LeRoutier> sri: perhaps it still points to sf.net CVS
[04:08] <sri> cvs -z3 -q -d :pserver:[email protected]:/cvs/gstreamer update -dP -A .
[04:08] <sri> is the command line from jhbuild.
[04:08] <sri> I think waht I will do is blow away everything related to gstreamer in my lib dir
[04:08] <Company> sri: clear your plugins dir
[04:08] <Company> sri: there might be old leftover plugins
[04:09] <sri> Company: okay, I think I did that yesterday and then re-installed..but I can try it again.
[04:09] <Company> hum
[04:09] <ds-work> run 'ldd gst-launch-0.7'.  it's probably picking up the wrong libgstreamer
[04:09] <sri> okay
[04:10] <sri> looks okay:libgstreamer-0.7.so.5 => /gnome2/lib/libgstreamer-0.7.so.5 (0x40018000)
[04:11] <sri> Company: I'm prettys ure I blew everything away..everything in $prefix/lib/gstreamer-0.7/ has a timestamp of yesterday
[04:12] <sri> okay, I think ds-work, you're correct, I'm going remove all the libgst* stuff in $prefix/lib and see what happens.
[04:33] Nick change: harshyOUT -> harshy
[04:37] drjimmy42 ([email protected]) joined #gstreamer.
[04:39] <drjimmy42> I just got gstreamer out of the freedesktop cvs archive.  It downloaded all these directories
[04:39] <drjimmy42> beethoven:~/engineering/gstreamer> ls
[04:39] <drjimmy42> CVS           common     gst-editor      gst-monkeyaudio  gst-record    www
[04:39] <drjimmy42> CVSROOT       cothreads  gst-guile       gst-player       gst-sandbox
[04:39] <drjimmy42> autoregen.sh  gst-all    gst-java        gst-plugins      gst-template
[04:39] <drjimmy42> basement      gst-bind   gst-media-test  gst-python       gstreamer
[04:39] <drjimmy42> I have tried several variations of ./autogen.sh in several directories, but they all try give me this
[04:39] <drjimmy42> + getting common from cvs
[04:39] <drjimmy42> cvs server: cannot find module `common' - ignored
[04:39] <drjimmy42> cvs [checkout aborted]: cannot expand modules
[04:39] <drjimmy42> There is something wrong with your source tree.
[04:39] <drjimmy42> You are missing common/gst-autogen.sh
[04:40] <drjimmy42> does anyone know what I'm doing wrong
[04:41] Company ([email protected]) left irc: Read error: 60 (Operation timed out)
[04:44] <LeRoutier> why taking all the CVS tree and not only needed modules ?
[04:44] <drjimmy42> I thought that co gstreamer would give me what I need.  It gave me the whole tree
[04:44] <drjimmy42> either way, I should be able to build what I need if I have everything.  Do you know what the problem is here?
[04:46] <LeRoutier> well, i have no clue of the problem you could have. 
[04:46] <LeRoutier> copy the exact line you type for checkout here plz 
[04:46] <drjimmy42> what do you do to get gstreamer?  ALl I really want to do is build rhythmbox
[04:46] Nick change: kh_out -> khalek
[04:47] <LeRoutier> well, i just do : emerge gstreamer gst-plugins (i have CVS ebuilds for my gentoo)
[04:47] <LeRoutier> but it is gentoo only
[04:47] <drjimmy42> I use gentoo as well, where do you get the cvs ebuilds?  breakmygentoo?
[04:48] <LeRoutier> nope, i did them myself
[04:48] <LeRoutier> give me your mail
[04:48] <LeRoutier> in PV
[04:57] <LeRoutier> good night everyone
[04:57] LeRoutier ([email protected]) left irc: "Leaving"
[04:58] Nick change: harshy -> harshyOUT
[05:10] dilinger ([email protected]) joined #gstreamer.
[05:30] drjimmy42 ([email protected]) left #gstreamer.
[06:12] walters ([email protected]) joined #gstreamer.
[06:21] ChrisHJW ([email protected]) joined #gstreamer.
[06:21] mxpxpod ([email protected]) joined #gstreamer.
[06:22] dennis_ ([email protected]) left irc: Read error: 113 (No route to host)
[06:23] ChrisHJW_log ([email protected]) left irc: Read error: 60 (Operation timed out)
[06:28] ChrisHJW_log ([email protected]) joined #gstreamer.
[06:36] sub_pop ([email protected]) left irc: "Client exiting"
[06:36] bluejay ([email protected]) joined #gstreamer.
[06:42] ChrisHJW ([email protected]) left irc: Read error: 110 (Connection timed out)
[06:46] sub_pop ([email protected]) joined #gstreamer.
[06:59] mxpxpod ([email protected]) left irc: "Guns kill people as much as spoons made Rosie O'Donnell fat.."
[07:01] mxpxpod ([email protected]) joined #gstreamer.
[07:25] Nick change: mxpxpod -> mx|sleep
[07:26] sri ([email protected]) left irc: "Client exiting"
[07:33] sri ([email protected]) joined #gstreamer.
[07:47] sri ([email protected]) left irc: "Client exiting"
[07:51] sri ([email protected]) joined #gstreamer.
[08:04] thaytoo ([email protected]) left irc: "leaving"
[08:48] Nick change: harshyOUT -> harshy
[08:57] kmaraas ([email protected]) left irc: "Leaving"
[09:04] afx ([email protected]) joined #gstreamer.
[09:05] afx ([email protected]) left irc: Client Quit
[09:08] afx ([email protected]) joined #gstreamer.
[09:11] Shoragan ([email protected]) joined #gstreamer.
[09:23] Nick change: bluejay -> bluejay|sleep
[09:23] walters ([email protected]) left irc: "out"
[09:27] thaytan ([email protected]) joined #gstreamer.
[09:38] swentel ([email protected]) joined #gstreamer.
[09:47] thaytan ([email protected]) left irc: Read error: 113 (No route to host)
[09:48] sublett ([email protected]) joined #gstreamer.
[09:48] markov ([email protected]) joined #gstreamer.
[09:48] thaytan ([email protected]) joined #gstreamer.
[09:58] harshy ([email protected]) left irc: "I quit for now"
[10:09] afx ([email protected]) left irc: "Client exiting"
[10:17] dolphy ([email protected]) joined #gstreamer.
[10:22] thomasvs ([email protected]) left irc: Read error: 110 (Connection timed out)
[10:26] <mathrick|sleep> yo dolphy
[10:26] <mathrick|sleep> dolphy: interested in some whining about gst-player?
[10:27] Nick change: mathrick|sleep -> mathrick
[10:29] Nick change: mathrick -> mathrick_
[10:29] Nick change: mathrick_ -> mathrick
[10:31] Nick change: mathrick -> mathrick_
[10:31] Nick change: mathrick_ -> mathrick
[10:32] <dolphy> mathrick: well tell me. but atm libgsplay is not the latest version. i m planning to commit a new pipeline soon.
[10:34] <mathrick> ** (gst-player:14528): WARNING **: specified empty bin "bin", not allowed
[10:34] <mathrick>  
[10:34] <mathrick> ** (gst-player:14528): WARNING **: Unrecoverable syntax error while parsing pipeline
[10:34] <mathrick> DEBUG: gstgconf: error parsing pipeline audio/x-raw-int ! identity ! { queue max-size-time=1000000000 ! osssink }
[10:34] <mathrick> no element "audio"
[10:34] <mathrick> GConf audio sink not found, using osssink
[10:34] <mathrick> (gst-player:14528): GStreamer-CRITICAL **: file gstinterface.c: line 140 (gst_implements_interface_cast): assertion `gst_element_implements_interface (GST_ELEMENT (from), iface_type)' failed
[10:34] <mathrick>  
[10:34] <mathrick> ** (gst-player:14528): CRITICAL **: file xoverlay.c: line 181 (gst_x_overlay_got_desired_size): assertion `GST_IS_X_OVERLAY (overlay)' failed
[10:34] <dolphy> mathrick: wow check your gconf keys dude :)
[10:34] <mathrick> dolphy: and then no video w/ avi, on mpeg it's totally unable to play
[10:34] <dolphy> mathrick: you probably keep using xvideosink
[10:35] <mathrick> dolphy: k, will check
[10:35] <dolphy> mathrick: and your audiosink looks broken too
[10:35] <mathrick> dolphy: audio is some custommade one w/ queue :)
[10:35] <mathrick> dolphy: but it should work, except there were some pb's with magic syntax, I have bug somewhere
[10:36] <mathrick> dolphy: audio/x-raw-int ! identity ! { queue max-size-time=1000000000 ! osssink } <-- my audiosink
[10:39] <mathrick> dolphy: http://bugzilla.gnome.org/show_bug.cgi?id=131720 , if you're interested
[10:40] jeroen_ ([email protected]) joined #gstreamer.
[10:43] markov ([email protected]) left irc: "bbl"
[10:45] <mathrick> dolphy: any idea what that means:
[10:45] <mathrick> <-- smb module init called -->
[10:45] <mathrick> LOCK
[10:45] <mathrick> added interface ip=10.0.0.2 bcast=10.0.0.255 nmask=255.255.255.0
[10:45] <mathrick> lp_load: refreshing parameters
[10:45] <mathrick> Initialising global parameters
[10:45] <mathrick> Attempting to register new charset UCS-2LE
[10:45] <mathrick> Registered charset UCS-2LE
[10:45] <mathrick> Attempting to register new charset UTF8
[10:45] <mathrick> Registered charset UTF8
[10:45] <mathrick> Attempting to register new charset ASCII
[10:45] <mathrick> Registered charset ASCII
[10:45] <mathrick> Attempting to register new charset 646
[10:45] <mathrick> Registered charset 646
[10:45] <mathrick> Attempting to register new charset ISO-8859-1
[10:45] <mathrick> Registered charset ISO-8859-1
[10:45] <mathrick> Attempting to register new charset UCS2-HEX
[10:45] <mathrick> Registered charset UCS2-HEX
[10:45] <mathrick> Substituting charset 'ISO-8859-2' for LOCALE
[10:45] Last message repeated 1 time(s).
[10:45] <mathrick> ?
[10:45] <mathrick> dolphy: it shows when I try to play mpeg
[10:46] thomasvs ([email protected]) joined #gstreamer.
[10:46] <mathrick> lo thomasvs
[10:48] <mathrick> dolphy: overall, avi's play ok, but on mpeg's it barfs badly
[10:50] <dolphy> mathrick: well maybe your ownmade pipeline is not parsed correctly as a bin by gst_gconf stuff
[10:51] <mathrick> dolphy: changed it to plain osssink, got rid of warnings, but it still trips over
[10:53] <mathrick> ** (gst-player:25127): WARNING **: couldn't set the right number of channels, enjoy the speed difference
[10:53] <mathrick> and that strange smb messages too, they puzzle me badly :)
[11:07] harshy ([email protected]) joined #gstreamer.
[11:11] sxpert_work ([email protected]) left irc: "Leaving"
[11:11] sxpert_work ([email protected]) joined #gstreamer.
[11:19] Rotty ([email protected]) joined #gstreamer.
[11:33] gheet ([email protected]) joined #gstreamer.
[12:19] swentel ([email protected]) left irc: 
[12:25] <thaytan> http://flowdesigner.sourceforge.net/home.html
[12:25] <thaytan> ^ these guys are using a flow based architecture too
[12:26] <thaytan> they only do pull scheduling though
[12:41] sack ([email protected]) left irc: "Client Exiting"
[12:52] Nick change: swenzzzz -> swentel
[12:58] alley_cat ([email protected]) joined #gstreamer.
[13:00] teuf ([email protected]) joined #gstreamer.
[13:00] <teuf> hi
[13:01] <teuf> any idea when a new 0.7 release will be done ?
[13:04] <thomasvs> teuf: I hope to get one done sometime in the next ten days
[13:04] <teuf> that would be cool :)
[13:06] <mathrick> thomasvs: what exactly do I need to build gst docs? It'd be nice to finally refresh that ancient 0.6 ref manual sitting in my devhelp ;)
[13:07] <thomasvs> mathrick: configure tells you what you're missing, no ?
[13:07] <thomasvs> I don't remember offhand, since I set it up once
[13:07] <thomasvs> but I remember making sure configure says what it's missing
[13:07] <mathrick> thomasvs: hmm, gonna look at that
[13:07] <thomasvs> basically, just a decent gtk-doc and xslt-proc toolchain
[13:08] <mathrick> thomasvs: IIRC, it spews hella lot of non well-formedness errors
[13:08] <mathrick> thomasvs: like, unknown entity &nbsp;
[13:08] <teuf> you need your xml calatalogs to be properly set up too, and the necessary stylesheets
[13:09] <thomasvs> mathrick: well, then I want to see bug reports :)
[13:09] <mathrick> teuf: what is 'properly'? few times i tried using xsltproc, it worked ok, and didn't seem to have any pb's with finding catalogs. But i'll double check
[13:10] <teuf> mathrick: properly == let the distro do it unless you are on debian
[13:10] <mathrick> teuf: I am on debian
[13:11] <dolphy> mathrick: hmm what files do you play then ?
[13:11] <teuf> mathrick: you can try http://www.daa.com.au/~james/files/build-xml-catalog-for-debian.sh
[13:13] <mathrick> dolphy: these are some short anime clips, 16MB & 12MBmpegs. Interested in me uploading them? I can't put them anywhere on the web, unfortunately
[13:13] <mathrick> teuf: ok, will try
[13:14] <dolphy> mathrick: are they on a smb share ?
[13:14] <mathrick> dolphy: hell no, local disk
[13:14] <mathrick> dolphy: i don't even have smb set up
[13:21] foser ([email protected]) joined #gstreamer.
[13:24] thomasvs ([email protected]) left irc: Remote closed the connection
[13:27] thomasvs ([email protected]) joined #gstreamer.
[13:31] Shoragan ([email protected]) left irc: "Leaving"
[13:42] wingo ([email protected]) joined #gstreamer.
[13:43] <mathrick> teuf: no go, seems I can't contact that host :\
[13:43] <thomasvs> wingo: dude
[13:43] sublett ([email protected]) left irc: "I like food, food is good!"
[13:43] <mathrick> teuf: is there any other place I can grab this script from?
[13:43] <thomasvs> wingo: they're going to thrash your "one-channel-float" setup in gstreamer if you're not careful :)
[13:44] <teuf> mathrick: doesn't work from here either, and I don't seem to have a local copy :-/
[13:44] Action: mathrick googles
[13:44] <mathrick> teuf: nvm me, I'm dumb today ;)
[13:55] <wingo> thomasvs: i don't know why I have to defend this every six months :P
[13:56] <wingo> also I don't know why company cares about int2float's complexity, it doesn't seem to be in his use areas
[13:57] <wingo> FWIW, I would like that plugin pair declared as part of my "territory" -- or of steveb or iain, someone with similar points of view ;)
[14:01] <thomasvs> wingo: I agree - but you need to assert it
[14:02] <thomasvs> wingo: having a document that explains it where I can point them to though would help
[14:02] <thomasvs> wingo: otherwise some people feel they can change anything they don't understand
[14:03] <thaytan> gst-launch-0.7 videotestsrc ! video/x-raw-yuv,width=321,height=241 ! ffcolorspace ! ximagesink suxors
[14:03] <thaytan> seems like ffmpeg doesn't handle odd widths too well
[14:04] <mathrick> thaytan: odd sizes are forbidden
[14:04] <thaytan> mathrick: so it seems, although there's nothing in code stopping it
[14:04] <mathrick> wingo: Company seems to be concerned more with autoplugging complexity rather than int2float itself. And ds shares his view, so you have both plugger guys against :)
[14:05] <mathrick> thaytan: AFAIK, it has something to do with [yuv] memory layout
[14:05] <mathrick> thaytan: or maybe it was xvimagesink, dunno, but definitely it was about mem layout
[14:06] <thaytan> probably that they're rounding their allocation of Cb Cr planes down rather than up
[14:06] <thaytan> whereas ffmpeg assumes the opposite
[14:06] <thomasvs> anyone using both ccache and distcc ?
[14:07] <wingo> mathrick: damn :P documentation it is, then, and maybe I can get a witch doctor to put a hex on those that change the float setup ;)
[14:07] iain ([email protected]) joined #gstreamer.
[14:11] <mathrick> wingo: _if_ you can convince Company & ds world will be a better place with multipad plugging, then probably they'll do it. But now both of them seem ready to kill for that case ;)
[14:11] <thaytan> ds: ds-work: awake?
[14:12] wheels ([email protected]) left irc: "using sirc version 2.211+KSIRC/1.3.9"
[14:13] <thaytan> ds: how tired were you when you committed adder?
[14:16] <iain> thaytan: did he manage to swap the src and sink pads round again?
[14:16] <thaytan> no, it doesn't compile at all :)
[14:16] ChrisHJW ([email protected]) joined #gstreamer.
[14:17] <iain> oh, thats easy to notice...swapping src and sinks round is trickier :)
[14:17] <thomasvs> thaytan: I'm fixing adder
[14:17] <thaytan> thomasvs: oh, I've fixed it here
[14:18] <thaytan> but feel free
[14:18] <thomasvs> thaytan: just finishing distcc and ccache integration here
[14:18] <thomasvs> thaytan: ah, ok
[14:18] <thaytan> I was just checking everything else builds before committing
[14:19] <thomasvs> well, he obviously didn't compile it did he ? :)
[14:20] <thomasvs> thaytan: mine's in, feel free to check if I missed something
[14:21] <thaytan> ok
[14:21] <thaytan> hey, it's fun to watch a DVD in two windows at once - one with and one without subtitles!
[14:22] <thaytan> pity about the audio sync 
[14:23] <thaytan> I've decided I'm not going to commit this DVD stuff until SOMEONE replies to my mail and comments on it :P
[14:23] <thomasvs> what do you want us to do man ? :)
[14:23] <thomasvs> you're the expert
[14:25] <thaytan> I'm hoping someone will say 'hey that's a crazy way to get the elements working together, you should have used a foopipe' ;)
[14:26] <thaytan> anyways, goodnight
[14:26] <mathrick> thaytan: if this makes you feel better, I can push my switch proposal on list too :). There seems to be some problems with that and tracks identification, but I think tag for that may be good
[14:26] <mathrick> thaytan: g'night
[14:27] <wingo> mathrick: can it be exempted from the autoplugging battle?
[14:27] <wingo> why would you use float audio anyway if you weren't doing pro audio?
[14:28] <wingo> hi iain
[14:28] <thomasvs> wingo: you're trying to convince the wrong people
[14:28] <thomasvs> wingo: mail on the list
[14:28] <wingo> night thaytan
[14:28] <wingo> i'm downloading my mail now, a few hundred squeezing down an african pipe
[14:28] <mathrick> wingo: thomasvs is right, i'm not yet writing autoplugger ;)
[14:29] <iain> hey wingo 
[14:35] Nick change: khalek -> kh_zZz
[14:57] sublett ([email protected]) joined #gstreamer.
[15:04] sub_pop ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[15:04] Shoragan ([email protected]) joined #gstreamer.
[15:04] sub_pop ([email protected]) joined #gstreamer.
[15:08] <wingo> grumpy mail sent to -devel ;)
[15:17] edmack ([email protected]) joined #gstreamer.
[15:17] LeRoutier ([email protected]) joined #gstreamer.
[15:17] <LeRoutier> hello
[15:17] <mathrick> lo LeRoutier
[15:17] <teuf> salut
[15:18] <edmack> Hi, I'm having a go at building a little gtk player with gst, and I was wondering, does gst-player use a thread bin, or a normal bin with a timeout or some other method to run gst_bin_iterate? gtk_timeout so far hasn't seemed very apt for me
[15:25] jeroen_ ([email protected]) left irc: Remote closed the connection
[15:26] <thomasvs> dolphy: wake up, edmack is asking you something
[15:27] <wingo> peace out kids.
[15:27] wingo ([email protected]) left irc: "I like core dumps"
[15:27] <dolphy> ah
[15:27] Shoragan_ ([email protected]) joined #gstreamer.
[15:27] Shoragan ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[15:28] <dolphy> edmack: gstplay uses a thread bin
[15:29] <edmack> ok, thanks
[15:29] <edmack> I'll go read up on it
[15:29] <dolphy> edmack: but still you have to put an idler that will iterate the gstplay object
[15:29] <edmack> ok
[15:30] <mathrick> thomasvs: where should I expect ref manual to be built / installed ?
[15:33] <thomasvs> mathrick: the api ref, in docs/gst/html
[15:33] <thomasvs> (built)
[15:34] <mathrick> thomasvs: hum, only handful of .png in there
[15:34] <thomasvs> mathrick: well, as I said - send me a configure/build log :)
[15:34] <edmack> What plugins are on the cool to do list?
[15:35] <mathrick> edmack: EQ for one, is being worked on
[15:35] <edmack> nifty
[15:36] <mathrick> thomasvs: BUILD_DOCS_FALSE='#' <-- does this mean they're not going to be built? OK, dcc'ing you config.log
[15:37] <thomasvs> mathrick: mail
[15:37] Action: thomasvs has never met anyone for whom dcc works
[15:37] <mathrick> thomasvs: ie. what exactly doesn't work with dcc?
[15:37] <LeRoutier> well, implementations sucks, won't work thru NAT/PAT, ...
[15:38] <thomasvs> mathrick: firewalling issues
[15:38] <mathrick> thomasvs: what adress? apestaart?
[15:39] <thomasvs> yep
[15:39] <thomasvs> bring it on
[15:39] <thomasvs> but, please
[15:39] <thomasvs> send me the output of configure and make, not config.log
[15:39] <mathrick> thomasvs: for me it worked few times, and i'm behind double NAT
[15:40] <mathrick> thomasvs: ie ./configure 2>&1 > configure.out ?
[15:42] <thomasvs> mathrick: yep
[15:42] <thomasvs> mathrick: and same for make
[15:43] mx|sleep ([email protected]) left irc: "Guns kill people as much as spoons made Rosie O'Donnell fat.."
[16:03] Uraeus ([email protected]) joined #gstreamer.
[16:07] <Uraeus> argh, there is nothing more annoying than overfriendly bank clerks
[16:12] <dolphy> Uraeus: hi :)
[16:13] <Uraeus> dolphy: ok, I am ordering my airplane tickets now, while two others hasn't replied yet I assume you have them under control ;)
[16:16] Nick change: LeRoutier -> LeRaway
[16:21] <dolphy> Uraeus: well so few things i have under CONTROL :)
[16:22] <alley_cat> anyone else get this? ERROR: from element /pipeline0/vorbisfile0: Internal GStreamer error: pad problem.  File a bug.
[16:26] hallibaby ([email protected]) joined #gstreamer.
[16:44] <mathrick> thomasvs: logs sent, btw
[16:45] <thomasvs> alley_cat: what does the debug message say that gets printed ?
[16:45] ChrisHJW ([email protected]) left irc: 
[16:46] <alley_cat> Additional debug info:
[16:46] <alley_cat> gstpad.c(2208):gst_pad_set_explicit_caps:
[16:46] <alley_cat> failed to negotiate (try_set_caps returned REFUSED)
[16:46] <alley_cat> ERROR       scheduler(24722) gstoptimalscheduler.c(2074):gst_opt_scheduler_iterate: [GstOptScheduler@0x8058660] in error state
[16:46] <thomasvs> alley_cat: and the pipeline is ?
[16:47] <alley_cat> filesrc location=file.ogg ! vorbisfile ! alsasink device=dsp1
[16:47] <alley_cat> same happens with any other audiosink or with mp3/mad
[16:48] <thomasvs> replace alsasink with fakesink
[16:49] <alley_cat> hmm, fakesink and ossink work now
[16:50] <thomasvs> I don't think device=dsp1 is a valid alsa string at all
[16:50] <thomasvs> but I don't know
[16:50] <alley_cat> odd, yesterday no sink worked, now only alsasink doesn't
[16:50] <alley_cat> it is
[16:50] <mathrick> alley_cat: device="hw:0" ?
[16:50] <alley_cat> dsp1 is defined in .asoundrc
[16:50] <alley_cat> worked fine until two days ago
[16:54] <alley_cat> seems like alsasink _needs_ an audioscale before it now
[16:54] <alley_cat> or rather it errors out now instead of staying silent with audio that doesn't have a 48khz samplerate
[16:58] BBB ([email protected]) joined #gstreamer.
[16:59] <mathrick> j0 888 ;)
[17:00] <BBB> howy
[17:00] <dolphy> BBB: heya :)
[17:01] <BBB> howdy
[17:09] <thomasvs> anyone remember what the OFFICIAL guide of mime types is ?
[17:09] <thomasvs> alley_cat: sounds logical, yeah
[17:09] <teuf> thomasvs: iana ?
[17:09] <thomasvs> teuf: yyeah, but I don't find the spec
[17:10] <teuf> http://www.iana.org/assignments/media-types/
[17:10] <thomasvs> thanks, that's the one
[17:14] trow ([email protected]) joined #gstreamer.
[17:14] Nick change: LeRaway -> LeRoutier
[17:21] <thomasvs> mathrick: your gstelement.sgml file has cvs conflicts
[17:22] <thomasvs> mathrick: you need to fix those first
[17:22] <thomasvs> mathrick: easiest is to just delete tmpl/*.sgml
[17:22] <thomasvs> hi trow
[17:22] <mathrick> thomasvs: ok, strange b/c I didn't touch these
[17:23] <thomasvs> mathrick: gtk-doc does
[17:23] <thomasvs> mathrick: it's annoying, but it happens
[17:23] <mathrick> thomasvs: k
[17:23] <thomasvs> mathrick: it happens more often with big API changes :)
[17:24] <thomasvs> ok, time to hack on the website some more
[17:25] <trow> hi thomasvs
[17:35] markov ([email protected]) joined #gstreamer.
[17:36] <mathrick> yo markov
[17:36] <markov> yoyo mathrick 
[17:36] <markov> g-boy :)
[17:42] teleyinex ([email protected]) joined #gstreamer.
[17:42] <teleyinex> hi
[17:42] <teleyinex> im trying to compile 0.7.3
[17:42] <teleyinex> and i get this error on docs:
[17:43] <teleyinex> *** Generating HTML output ***
[17:43] <teleyinex> <teleyinex> Usage: jw [<options>] <sgml_file>
[17:43] <teleyinex> ........
[17:43] <teleyinex>  make[4]: *** [html/index.html] Error 1
[17:43] <teleyinex> make[4]: Leaving directory `/home/daniel/garnome-0.23.1/gnome/gstreamer/work/main.d/gstreamer-0.7.3/docs/faq'
[17:43] <teleyinex> what could be?
[17:43] <teleyinex> and why i get this error?
[17:43] <thomasvs> teleyinex: did you run make only in the toplevel dir ?
[17:43] <thomasvs> teleyinex: file a bug report with COMPLETE output of configure and make so I can check
[17:44] <teleyinex> in #flood channel?
[17:44] ChrisHJW ([email protected]) joined #gstreamer.
[17:44] <thomasvs> teleyinex: no, file a bug report in bugzilla so I can see it easily
[17:45] <teleyinex> where?
[17:45] <teleyinex> in what page?
[17:45] <teleyinex> change page, for site
[17:45] <thomasvs> bugzilla.gnome.org
[17:46] <LeRoutier> product : Gstreamer, element : core, version : 0.7.3
[17:57] <BBB> hm
[17:57] <BBB> more fixing tonight
[17:57] Action: BBB goes have food now
[17:57] <BBB> bye
[17:57] BBB ([email protected]) left irc: "Client exiting"
[17:57] markov ([email protected]) left irc: "leaving"
[17:58] Nick change: LeRoutier -> LeRaway
[18:08] <teleyinex> thomasvs, what you want about make?
[18:09] <thomasvs> teleyinex: both ./configure > configure.out 2>&1 and make > make.out 2>&1
[18:14] <teleyinex> could be a problem, that im using garnome?
[18:19] <thomasvs> teleyinex: I have no idea, and I'll tell you when I have the logs
[18:38] hallibaby_ ([email protected]) joined #gstreamer.
[18:41] Uraeus ([email protected]) left irc: "Client exiting"
[18:43] <thomasvs> hm.
[18:43] <thomasvs> it doesn't look like I can get the mime type easily from caps anymore
[18:45] teuf ([email protected]) left irc: "Client exiting"
[18:46] Nick change: LeRaway -> LeRoutier
[18:48] <teleyinex> well
[18:48] edmack ([email protected]) left #gstreamer.
[18:49] <teleyinex> now i paste the files there, or i send you the files?
[18:49] <teleyinex> cause in bugzilla.gnome.org i dont see nothing to upload files
[18:49] <thomasvs> teleyinex: you can mail me at thomas at apestaart dot org
[18:49] <thomasvs> teleyinex: or you can attach a file in bugzilla
[18:49] <thomasvs> teleyinex: there is, "create attachment"
[18:52] <teleyinex> sorry but i dont see the option create attachment
[18:52] <teleyinex> im at bugzilla.gnome.org
[18:52] <teleyinex> there, i select The traditional and more sophisticated bug interface is available for people already familiar with bugzilla.
[18:52] <teleyinex> then i select the product: Gstreamer
[18:53] <teleyinex> and in this page, where i select: version, OS, etc... i get stucked
[18:54] hallibaby ([email protected]) left irc: Read error: 110 (Connection timed out)
[18:55] <teleyinex> i have send you an email
[18:55] <teleyinex> with the two outputs
[19:00] <thomasvs> ok, I'll wait for it
[19:00] <mathrick> teleyinex: to create attachment, you need to create bug first, and only then click "create attachment" in it
[19:00] <thomasvs> teleyinex: but, after creating a bug, you can attach
[19:03] Misirlou ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[19:05] Misirlou ([email protected]) joined #gstreamer.
[19:08] gheet ([email protected]) left irc: "Download Gaim [http://gaim.sourceforge.net/]"
[19:14] <teleyinex> ok
[19:14] <teleyinex> well
[19:15] <teleyinex> do you want that i do that? or with the mail isnt necessary
[19:15] <thomasvs> I haven't gotten your mail yet, so I'd appreciate it being in bugzilla yeah
[19:19] Action: mathrick reboots to test 2.6.1
[19:20] mathrick ([email protected]) left irc: Remote closed the connection
[19:23] <teleyinex> ok
[19:23] <teleyinex> tomorrow i will do
[19:23] <teleyinex> cya
[19:23] <teleyinex> and thanks for all
[19:23] teleyinex ([email protected]) left irc: "Terminando cliente"
[19:37] markov ([email protected]) joined #gstreamer.
[19:40] <taaz> i leave for a few days and you all send out 1000 emails.  grrr! ;)
[19:40] <thomasvs> taaz: yeah - and eight about DVD's !
[19:46] mathrick ([email protected]) joined #gstreamer.
[19:47] Action: mathrick failed at running 2.6 for first time
[19:53] jdahlin ([email protected]) joined #gstreamer.
[19:54] murrayc_ ([email protected]) joined #gstreamer.
[19:54] <murrayc_> This could be my fault, I suppose, but I just built gstreamer, gst-plugins, and rhythmbox, and I get this when starting rhythmbox:
[19:54] <murrayc_> (rhythmbox:10213): GLib-GObject-WARNING **: specified class size for type `GstAutoplugger' is smaller than the parent type's `GstBin' class size
[19:54] <murrayc_>  
[19:54] <murrayc_> (rhythmbox:10213): GStreamer-CRITICAL **: file gstelementfactory.c: line 188 (gst_element_register): assertion `g_type_is_a (type, GST_TYPE_ELEMENT)' failed
[19:54] <murrayc_> rhythmbox: relocation error: /opt/gnome26/lib/gstreamer-0.7/libgststaticautoplugrender.so: undefined symbol: gst_autoplug_get_type
[19:55] <thomasvs> murrayc_: you need to rebulid core
[19:55] <murrayc_> gstreamer?
[19:55] <thomasvs> murrayc_: you have both old symbols, and plugins with a different ABI size
[19:56] <murrayc_> Time to delete everything and get a new checkout then. I agree, the size thing does seem illogical if it built properly.
[19:59] KoRnouille ([email protected]) joined #gstreamer.
[20:16] hallibaby_ ([email protected]) left irc: Remote closed the connection
[20:22] <LeRoutier> is there any work related to MIDI done somewhere for gstreamer ?
[20:23] <LeRoutier> (i just discovered that karaoke files like .kar are in fact some MIDI files with text "stream" inside
[20:23] hallibaby ([email protected]) joined #gstreamer.
[20:23] thomasvs ([email protected]) left irc: No route to host
[20:27] <mathrick> LeRoutier: docs/random/uraeus/gstreamer_and_midi.txt
[20:27] <LeRoutier> k, thx
[20:29] <murrayc_> Very strange. I deleted gstreamer and gst-plugins, rebuilt them with jhbuild and I get the same error.
[20:29] shanatk ([email protected]) joined #gstreamer.
[20:29] shanatk ([email protected]) left irc: Client Quit
[20:30] <LeRoutier> are you sure you haven't got another gstreamer 0.7.x somewhere else in another path ?
[20:30] shanatk ([email protected]) joined #gstreamer.
[20:30] <murrayc_> LeRoutier: Probably in /usr, because this is RedHat, but I don't know why it would be getting it.
[20:31] <mathrick> murrayc_: find / -iname "*gst*"
[20:31] shanatk ([email protected]) left #gstreamer.
[20:31] shanatk ([email protected]) joined #gstreamer.
[20:31] <mathrick> murrayc_: if that shows nothing, then it's strange indeed
[20:31] <shanatk> hello world
[20:31] <mathrick> hello shanatk
[20:31] <murrayc_> What's -iname?
[20:32] <mathrick> murrayc_: case insensitive
[20:32] <murrayc_> Well of course that's going to find something.
[20:32] shanatk ([email protected]) left irc: Client Quit
[20:32] <murrayc_> Don't worry, it must just be a problem on my system.
[20:32] jdahlin ([email protected]) left #gstreamer ("Leaving").
[20:33] shanatk ([email protected]) joined #gstreamer.
[20:34] murrayc_ ([email protected]) left irc: "Client Exiting"
[20:35] <shanatk> does anyone know the status of the int2float plugin ?
[20:39] KoRnouille ([email protected]) left irc: "Leaving"
[20:40] shanatk ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[20:41] shanatk ([email protected]) joined #gstreamer.
[20:45] <ds-work> shanatk: it works
[20:49] kmaraas ([email protected]) joined #gstreamer.
[20:50] Uraeus ([email protected]) joined #gstreamer.
[20:50] sub_pop ([email protected]) left irc: "Client exiting"
[20:51] <dolphy> ds-work: were you able to take the time too look at my pipeline issue?
[20:52] <dolphy> ds-work: s/too/to
[20:52] <dolphy> ds-work: i think it was filesrc location="matrix.avi" ! spider ! goom ! ffcolorspace ! videoscale ! ximagesink
[20:53] <dolphy> ds-work: having videoscale apparently breaks goom negotiation and the colors are very strange
[20:53] <dolphy> bbl
[20:53] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[20:53] <shanatk> ds-work: not with the identity plugin!!
[20:54] thomasvs ([email protected]) joined #gstreamer.
[20:55] <Uraeus> hmm, what is happening our mailing list is suddenly high traffic :)
[20:57] <LeRoutier> OSNews article effect N
[20:57] <LeRoutier> ?
[20:57] Action: markov can't get identity to work properly :/
[20:59] KoRnouille ([email protected]) joined #gstreamer.
[21:04] Nick change: bluejay|sleep -> bluejay
[21:07] <LeRoutier> hum, too bad. mplayer does not play my .kar files : LMLM4 Stream Format not found
[21:10] KoRnouille ([email protected]) left irc: "Leaving"
[21:23] <LeRoutier> seems it can't play midi files at all as trying to play a plain .mid would throw the same error
[21:29] BBB ([email protected]) joined #gstreamer.
[21:29] <LeRoutier> hello BBB
[21:30] <BBB> hi
[21:30] <markov> re BBB 
[21:43] Action: BBB is fixing mixer bug
[21:43] Action: BBB is getting annoyed by the horrendously low internet speed here
[21:44] Action: BBB gives up
[21:44] <BBB> I'll commit my stuff tomorrow
[21:45] Nick change: swentel -> swenzzzz
[21:46] Zeenix ([email protected]) joined #gstreamer.
[21:46] <Zeenix> hello
[21:47] Action: Zeenix is still using windoz.. :(
[21:48] <BBB> geh :p
[21:59] <markov> heh
[22:01] <BBB> more bugs? :)
[22:01] Action: BBB awaits
[22:01] <BBB> thomasvs: what media formats need love?
[22:03] Action: ds-work feels stupid
[22:03] Action: ds-work left the laptop on at home, but with the ethernet cable disconnected
[22:03] <BBB> :D
[22:04] <ds-work> today would be a good day to get wireless working
[22:05] <BBB> does wavparse work ok now? did you test it?
[22:09] <ds-work> i didn't know there were problems
[22:10] sub_pop ([email protected]) joined #gstreamer.
[22:14] Company ([email protected]) joined #gstreamer.
[22:18] apoc ([email protected]) joined #gstreamer.
[22:19] <apoc> yo
[22:20] ajmitch ([email protected]) joined #gstreamer.
[22:22] <apoc> hey ajmitch
[22:22] wheels ([email protected]) joined #gstreamer.
[22:22] <ajmitch> hello
[22:22] kh_zZz ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[22:25] <BBB> ds-work: I sent you a file yesterday!
[22:25] <BBB> ds-work: filesrc location=file.mp3inwav ! wavparse ! spider ! osssink
[22:26] <BBB> wavparse detects the type, sets explicit caps correctly
[22:26] <BBB> but somehow, things still go wrong
[22:26] <ds-work> BBB: it's a spider bug
[22:26] dolphy ([email protected]) joined #gstreamer.
[22:26] <BBB> gr
[22:26] <BBB> crap
[22:26] <dolphy> re
[22:26] <BBB> so how do we fix it?
[22:26] <BBB> will the new autoplugger be finished before 0.8.0?
[22:26] <ds-work> BBB: yes
[22:27] <Company> it's supposed to
[22:27] <ds-work> BBB: at the very least, there will be a static autoplugger
[22:27] <Company> and wavparse ! spider ! osssink doesn'T work because spider plugs them directly
[22:27] <BBB> ok
[22:27] <BBB> I'll wait then
[22:28] <Company> you could "fix" it like every other element byy making it use sometimes pads...
[22:28] m_wheels ([email protected]) joined #gstreamer.
[22:28] wheels ([email protected]) left irc: Nick collision from services.
[22:29] Nick change: m_wheels -> wheels
[22:29] <BBB> Company: well, wavparse needs that anywya
[22:29] <BBB> it can contain multiple audio tracks
[22:29] <Company> but it contains at least one...
[22:29] <BBB> but I'll do that as i rewrite it to work with rifflib
[22:30] <BBB> yes...
[22:30] <BBB> well, ALWAYS_1, sometimes_more
[22:30] <BBB> whatever
[22:30] <dolphy> ds-work: re.. any ideas on the pipeline i gave you?
[22:31] <BBB> Company: is there a mimetype doc somewhere in docs/random?
[22:31] <BBB> Company: er
[22:31] <BBB> tagging
[22:31] <BBB> Company: is there a tagging doc somewhere in docs/random?
[22:31] <ds-work> dolphy: what was the link again?
[22:32] Action: BBB wants to make avi/matroska use the new tag system
[22:32] <Company> BBB: i think ds put it in docs/random/company
[22:32] <BBB> cool!
[22:33] <dolphy> ds-work: filesrc location="/home/dolphy/matrix.avi" ! spider ! goom ! ffcolorspace ! videoscale ! ximagesink
[22:33] <dolphy> ds-work: that one is broken
[22:33] <dolphy> ds-work: that one works
[22:34] <dolphy> ds-work: filesrc location="/home/dolphy/matrix.avi" ! spider ! goom ! ffcolorspace ! ximagesink
[22:34] <ds-work> dolphy: yes, it should be fixed
[22:34] <BBB> videoscale needs to be before ffcolorspace, afaik
[22:34] <BBB> (for now)
[22:34] <BBB> bl tomorrow
[22:34] <BBB> too expensive like this
[22:34] <BBB> bye
[22:34] BBB ([email protected]) left irc: "Client exiting"
[22:35] <ds-work> dolphy: among other bug fixes, I added a fixate function to goom
[22:36] <dolphy> ds-work: ok i will try that
[22:36] <ds-work> dolphy: but in the case of goom!...!ximagesink, you shouldn't use videoscale
[22:37] <dolphy> ds-work: well it s part of the gstplay pipeline you know..
[22:37] <ds-work> dolphy: also, you must use app fixate functions if you actually care about the format of the stream
[22:37] <dolphy> ds-work: there's colorbalance and so on
[22:38] <Company> videoscale should try to set incoming caps on the other side in the link function and if that fails scale itself
[22:38] <ds-work> it does
[22:38] <Company> in that case videoscale shouldn'T be a problem
[22:39] Action: dolphy nods
[22:39] <Company> in a goom ! videoscale ! ximagesink pipeline
[22:39] <dolphy> videoscale should be passthru
[22:43] jimmy_dean_ ([email protected]) joined #gstreamer.
[22:45] jimmy_dean ([email protected]) left irc: Read error: 110 (Connection timed out)
[22:47] <dolphy> ds-work: will synaesthesia need the same fixes ?
[22:48] <ds-work> dolphy: potentially
[22:48] thaytan ([email protected]) left irc: No route to host
[22:49] <ds-work> synasthesia is pretty broken wrt negotiation right now
[22:53] <dolphy> ds-work: an element going from READY to NULL might come back to READY ?
[22:54] <ds-work> yes
[22:54] <dolphy> ds-work: hmm and it will ALWAYS go to NULL before being disposed ?
[22:55] <dolphy> ds-work: because actually x[v]imagesink are freeing their xcontext on dispose
[22:55] <dolphy> ds-work: not on READY_TO_NULL
[22:55] <dolphy> ds-work: and i m wondering if i should change that
[22:55] ct_ ([email protected]) joined #gstreamer.
[22:56] <ds-work> should in in READY->NULL
[22:56] <dolphy> ok
[22:56] <ds-work> since it could theoretically open a new connection
[22:56] Action: dolphy moves the code from dispose to statechange
[22:57] <ds-work> cool.  that means we can move video windows to a new display without pausing :)
[22:58] shanatk ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[23:00] ChrisHJW ([email protected]) left irc: Connection timed out
[23:02] <Company> arguably changing the display property should do this without state changes
[23:02] <Company> the same for lopcation on filesrc
[23:03] <ds-work> no
[23:03] <dolphy> ds-work: going to NULL without PAUSED ?
[23:04] <dolphy> ds-work: you plan to do that ? :)
[23:04] <ds-work> you have to switch to an appropriate state for some property changes to take effect
[23:04] <Company> and why would location changes need that?
[23:04] <ds-work> dolphy: no, but you can change ximagesink without changing the rest of the pipeline
[23:04] <dolphy> ds-work: true
[23:04] <ds-work> Company: location changes need to go to READY
[23:05] <Company> if someone sets the property, shouldn't the property just change?
[23:05] <dolphy> ds-work: i m also adding a flag telling if the colorbalance settings have been changed in xvimagesink before the xcontext is initialized
[23:05] <ds-work> Company: no
[23:05] <Company> it's dead simple actually
[23:05] <Company> why not?
[23:05] <ds-work> Company: the property changes, but it doesn't take effect
[23:05] <dolphy> ds-work: if they haven't been set then getting the xcontext gather the value from the Xv port as defaults
[23:05] <Company> ds-work: that is really messy - what does get_property return then?
[23:06] <dolphy> ds-work: some buggy Xv drivers set their default not in the middle of the range :-[
[23:06] <ds-work> Company: get_property returns void
[23:06] <Company> void?
[23:06] <Company> get_property cannot return void - it can return NULL though
[23:07] <ds-work> ah
[23:07] <Company> but that would break gst-editor badly
[23:07] <ds-work> the property appears to have changed, but the change doesn't take effect
[23:07] <ds-work> this is how it's always worked
[23:07] <Company> so there's actually no way to figure out what file is playing?
[23:07] <ds-work> because you can't deny a property change
[23:08] <Company> you can deny a property change, we do it in lots of places - just don't change the property
[23:08] Uraeus ([email protected]) left irc: "Client exiting"
[23:08] <Company> to know if a property really changed you need to listen to notifies or use get_property afterwards
[23:09] <Company> at least that's how i thought it works...
[23:09] <ds-work> in glib, you're not supposed to deny property changes
[23:09] <dolphy> Company: you just made me think i should g_object_notify on properties change of x[v]imagesink
[23:10] <Company> ds-work: uhm, but that doesn't work with our code...
[23:11] <ds-work> dolphy: any reason colorbalance uses [-1000, 1000] instead of doubles?
[23:11] <ds-work> Company: wait for glib-3.0
[23:12] <Company> ds-work: but seriously, this means that get_property for filesrc might not return the currently playing file?
[23:12] <ds-work> Company: there's no way to signal that you can't open the new file if you set the location in filesink while playing
[23:12] <ds-work> Company: you are correct
[23:12] <Company> ds-work: just don't set it and don't g_object_notify?
[23:13] <ds-work> Company: and then what state should you be in?  continue to use the old file?
[23:13] <Company> ds-work: yeah, just behave as if the function call failed
[23:14] <ds-work> GStreamer doesn't currently work like that
[23:14] <Company> i think it does
[23:15] <ds-work> I can dig up IRC logs with wim and omega that says it doesn't
[23:15] <Company> i can open up gstfilesrc.c and see it does
[23:16] <ds-work>   /* the element must be stopped in order to do this */
[23:16] <ds-work>   if (GST_STATE (src) == GST_STATE_PLAYING)
[23:16] <ds-work>     return FALSE;
[23:16] <Company> yeah
[23:17] <Company> so a set_property can fail :)
[23:17] <ds-work> you wrote that code
[23:18] <Company> i moved that
[23:18] <Company> and i didn't write the code for the mapsize argument
[23:20] <Company> i used that method in all elements i coded btw
[23:21] <ds-work> why don't all set_property functions use g_object_notify()?
[23:21] <Company> because noone added it
[23:21] <dolphy> ds-work: colorbalance ?
[23:22] <dolphy> ds-work: you mean xvimagesink properties ?
[23:22] <ds-work> dolphy: that, too
[23:22] <Company> i dunno if glib does a notify automagically (iain said that some day)
[23:23] <dolphy> ds-work: well the reason why xvimagesink has -1000 1000 on the properties is that according to XVideo doc that's the range of Xv settings
[23:23] <dolphy> ds-work: some drivers are not respecting that though
[23:23] <dolphy> ds-work: so i have to map with runtimes ranges
[23:24] <ds-work> dolphy: I'd prefer to have a specific mapping from properties to Xv settings
[23:24] <ds-work> dolphy: instead of adjusting to a particular range
[23:24] <dolphy> ds-work: there is such a mapping
[23:24] <dolphy> ds-work: i m not adjusting
[23:25] <dolphy> ds-work: i had to choose a range for properties
[23:25] <Company> hm yeah
[23:25] <Company> g_object_set does a notify automagically
[23:25] <ds-work> Company: gtk doesn't call g_object_notify()
[23:26] <Company> but that doesn't matter anyway
[23:26] <Company> because you'll just return the old parameter if you failed anyway
[23:26] <ds-work> Company: I don't really care about the behavior with set_property().  I initially thought you were saying that elements should allow property changes in states that don't make sense
[23:27] <Company> i would like them to allow that
[23:27] <Company> but if you think it's not good to change filesrc's location in PLAYING, i'd be ok with that
[23:27] <ds-work> changing the ximagesink:display property in PLAYING doesn't make sense
[23:28] <Company> why not? as an app developer i'd assume the ximagesink to move to another display then
[23:28] <ds-work> because we have state changes for that
[23:28] <dolphy> ds-work: filesrc ! spider ! goom ! ffcolorspace ! videoscale ! ximagesink still fails
[23:28] <dolphy> ds-work: filesrc ! spider ! goom ! videoscale ! ffcolorspace !  ximagesink works though
[23:29] <Company> do we want to force apps to do state changes?
[23:29] <ds-work> dolphy: it works fine here.
[23:30] <Company> especially because apps don't know in which state changing a particular proerty might be allowed
[23:31] <ds-work> Company: that's what documentation is for
[23:31] <Company> yeah, if you think that's better we do it that way, i don't really care
[23:32] <Company> though it would be cool if you could do runtime inspection of what needs to be done to make a property settable
[23:32] <ds-work> Company: on the other hand, if you change the property, then do a PLAYING->PAUSED->PLAYING, the change to PLAYING will fail
[23:32] <ds-work> I agree
[23:32] <Company> filesrc's location change must be done in NULL
[23:33] <Company> that's a bug in filesrc...
[23:33] <dolphy> ds-work: have you commited everything ?
[23:33] <ds-work> I think that makes more sense than an element doing an implicit state change to set a property
[23:34] <ds-work> Company: we could add a "property failed to set" in GstElement
[23:34] <ds-work> that would not suck
[23:34] <Company> apart from not knowing what you need to do as an app developer i agree there
[23:34] <ds-work> dolphy: yes.  perhaps you're complaing about the 16x16 window?
[23:35] jdahlin ([email protected]) joined #gstreamer.
[23:36] <dolphy> ds-work: and what's inside aswell
[23:36] <jdahlin> evening dolphy
[23:36] <dolphy> ds-work: putting videoscale next to videobalance works
[23:36] <dolphy> jdahlin: hey :)
[23:36] <dolphy> ds-work: should i let it there ?
[23:38] <ds-work> dolphy: you need to somehow tell the pipeline what size and framerate to use, because the format is underspecified in both pipelines
[23:38] <ds-work> in one pipeline, it gets a format that you like
[23:38] <ds-work> in the other, it doesn't
[23:38] <ds-work> in both cases, it's guessing
[23:39] <dolphy> ds-work: but when i m resizing the video window it should scale
[23:39] <dolphy> ds-work: but i get a black window
[23:40] <dolphy> ds-work: http://getcontrol.sourceforge.net/files/gstplay_pipeline.png can i use that one safely (please focus on the video thread) ?
[23:42] <ds-work> dolphy: it probably is using a framerate of 1.0
[23:42] <dolphy> ds-work: which is why i see nothing ?
[23:42] <ds-work> gst-launch-0.7 filesrc location=/home/ds/music/spaceghost.theme.mp3 ! spider ! goom ! ffcolorspace ! videoscale ! video/x-raw-rgb, framerate=30.0 ! ximagesin
[23:42] <ds-work> k
[23:43] <dolphy> ds-work: works
[23:43] <dolphy> ds-work: ok i understand now
[23:43] <dolphy> ds-work: so which combination is best ?
[23:44] <thomasvs> booboo
[23:44] <dolphy> ds-work: videobalance ! ffcolorspace ! videoscale ! ximagesink or videobalance ! videoscale ! ffcolorspace ! ximagesink ?
[23:44] <ds-work> dolphy: it doesn't matter
[23:44] <ds-work> dolphy: if you're enlarging, probably ffcolorspace ! videoscale
[23:45] <ds-work> and reducing isn't really that interesting, imo
[23:45] <dolphy> ds-work: hmm that's the one who needs to be enforced at framerate 30
[23:46] <ds-work> dolphy: you should probably set that at goom's src pad
[23:46] <dolphy> ds-work: in fixate ?
[23:46] <ds-work> dolphy: in this case, it's not necessary.  filter caps work fine
[23:47] <dolphy> ds-work: goom is already fixating at 30
[23:47] <dolphy> ds-work: who is fixating at 1 then ?
[23:47] <ds-work> ok, I'm going to repeat this one more time
[23:47] <ds-work> FIXATING IS GUESSING
[23:47] <dolphy> ds-work: sorry but i m learning slowly :)
[23:48] <thomasvs> ds-work: is there a mechanism to correct the guessing ?
[23:48] <ds-work> thomasvs: connect to a pad's fixate signal
[23:48] <ds-work> preferably all of them
[23:48] <dolphy> ds-work: i don't know much about filtered caps. connecting videoscale to ximagesink with filtered caps setting framerate to 30 will not break video playback from some file with a different framerate ?
[23:49] <thomasvs> ds-work: can you expand on that a little so I follow ? :)
[23:49] <thomasvs> ds-work: think, for example, the mad problem I posted to the list
[23:50] <ds-work> thomasvs: the mad problem doesn't matter, since both elements ignore the mpegversion field
[23:50] <ds-work> dolphy: set the filtered caps right after goom
[23:51] <thomasvs> ds-work: right, but I was using the nego'd layer in media'info
[23:51] <thomasvs> ds-work: so, there's no way to make sure that the right layer is nego'd ?
[23:51] <ds-work> as in,gst-launch-0.7 filesrc location=/home/ds/music/spaceghost.theme.mp3 ! spider ! goom ! video/x-raw-rgb, framerate=30.0\; video/x-raw-yuv, framerate=30.0 ! ffcolorspace ! videoscale ! ximagesink
[23:52] <dolphy> ds-work: ok seems wise
[23:52] Action: dolphy now looks how to make filtered caps :)
[23:52] <ds-work> thomasvs: arguably, after seeing its stream, mad could change the sink caps
[23:53] <ds-work> thomasvs: or you could use typefind
[23:54] trow ([email protected]) left irc: "Leaving"
[23:54] <LeRoutier> gst-typefind now gives a different "mime-type" for MP3s with and without ID3 tags ?!
[23:56] markov_ ([email protected]) joined #gstreamer.
[23:56] <Company> LeRoutier: yes
[23:56] markov ([email protected]) left irc: "Lost terminal"
[23:56] <LeRoutier> application/x-id3 vs audio/mpeg, mpegversion=(int)1|2, layer=(int)3
[23:56] <Company> LeRoutier: we defined id3 tagging as a container format
[23:57] apoc ([email protected]) left irc: "Leaving"
[23:57] Nick change: markov_ -> markov
[23:57] <Company> <ds-work> thomasvs: connect to a pad's fixate signal
[23:57] <Company> <ds-work> preferably all of them
[23:57] <Company> that's hard to do
[23:57] <thomasvs> ds-work: ok - but basically since fixating is guessing, and it needs to be done before data can be flow, we have no way to guess the right way anymore, right ?
[23:57] <Company> you'll never know which fixating function might be triggered...
[23:57] <thomasvs> ds-work: not complaining, just checking if I understand the problem
[23:57] <thomasvs> ds-work: I'm trying to figure out how an app should figure out if it's layer 1, 2, or 3
[23:58] <Company> yes, if multiple caps are possible, one of the possibilities is taken at random
[23:58] <ds-work> Company: exactly.  so you need to connect to all of them to make sure that you're called with the important fixate call happens
[23:58] <Company> so in short: fixating sucks, use filtered caps :)
[23:58] <jdahlin> thomasvs, is #114216 fixed?
[23:58] Action: thomasvs is confused what change in the caps system caused something like layer detection to not be possible anymore
[23:59] <Company> thomasvs: use a plugin that does layer detection
[23:59] <thomasvs> jdahlin: yeah, I'll close it
[23:59] <Company> thomasvs: (btw, layer did not work via caps in 0.6)
[23:59] <thomasvs> Company: but you can't before actually looking at the data, can you ?
[00:00] --- Fri Jan 23 2004
[00:00] <thomasvs> how does it work for video ?
[00:00] <thomasvs> framerate is part of caps, no ?
[00:00] <Company> yes, you can'T before looking at the data
[00:00] <thomasvs> if you read an mpeg file, where does the framerate get settled ?
[00:00] <Company> framerate is set after looking at the data
[00:00] <thomasvs> but it's part of the caps ?
[00:00] <Company> yes
[00:00] <ds-work> thomasvs: it just happened to guess correctly in 0.6
[00:00] <thomasvs> ok, so why can't mad do the same ?
[00:00] <Company> mad does that
[00:01] <Company> internally
[00:01] <thomasvs> ok :) but I'm trying to figure out why it's possible for video and framerate, but not for audio and layer
[00:01] <Company> theres no need for correct caps as mad doesn't export that layer info via caps
[00:01] <ds-work> thomasvs: it could, it just doesn't set the sink pad caps once it knows
[00:01] <thomasvs> ah, ok
[00:01] <thomasvs> so in the framerate case, it also fixates, but on a possibly wrong framerate
[00:01] <thomasvs> than later corrects it  ?
[00:01] <Company> no
[00:02] <ds-work> no, it delays the link
[00:02] <Company> it oesn't fixate until it knows
[00:02] <thomasvs> ah
[00:02] <thomasvs> so you're saying, while mad COULD delay, it's not worth it ?
[00:02] <Company> in the framerate case the src side decides the framerate
[00:02] <Company> that doesn't work in the filesrc ! mad case
[00:02] <thomasvs> Company: which src is that in the framerate case ?
[00:02] <Company> mpegdemux
[00:03] <ds-work> no, mad _could_ change the already negotiated caps on its sink pad just so that it can have the caps exactly correct
[00:03] <thomasvs> ok
[00:03] thaytoo ([email protected]) joined #gstreamer.
[00:03] <thomasvs> ds-work: so which is best - remove layer from the caps completely, or have it fix it afterwards ?
[00:03] <Company> neither
[00:03] <ds-work> layer is necessary for encoding
[00:03] <Company> use an element for detection if you care
[00:04] <Company> mpegparse should do it for example
[00:04] <thomasvs> yes, but from mad's caps, we can remove it without problem no ?
[00:04] <Company> or is it mp3parse?
[00:04] <Company> from mad's sinkcaps?
[00:04] <thomasvs> there's no reason to keep it in if it doesn't get used and doesn't get detected properly.  or am I missing something ?
[00:04] <Company> dunno
[00:04] <ds-work> thomasvs: probably.  but optional fields are challenging
[00:05] <Company> if someone invents a layer 4 it would be needed, but not now :)
[00:05] <thomasvs> ds-work: there's no way that fixating caps could "taint" struct members that have been guessed randomly ?
[00:05] <ds-work> thomasvs: huh?
[00:05] <Company> ds-work: optional fileds need to autofixate to a "none" value
[00:05] <Company> ds-work: so mad should advertise layer=[0,3]
[00:05] <thomasvs> ds-work: well, if it would be clear from the fixed caps that the layer was just a guess or not valid, I could at least notice something went wrong
[00:06] <Company> with 0 being "not set"
[00:06] <thomasvs> Company: yeah, exactly
[00:06] <thomasvs> fixated, but not able to decide on what
[00:06] <ds-work> Company: sure, but that's about 5th on the list of things needed to support optional fields
[00:06] <Company> that's what I want to do for all optional fields
[00:06] <Company> like framerate
[00:06] <thomasvs> I'd rather have it tell me it's layer 0 than layer 1 :)
[00:06] <Company> 0.0 means not set
[00:07] <ds-work> thomasvs: call gst_pad_get_caps() on the pad
[00:07] <Company> but i haven't thought that one through so i don't know if it'd work good enough ain all the cases
[00:07] <ds-work> thomasvs: the format could be anything that is returned
[00:07] <thomasvs> ds-work: I don't follow - call that to do what ?
[00:08] <ds-work> Company: that's a hack
[00:08] <Company> ds-work: but one that would work :)
[00:08] <ds-work> thomasvs: call gst_pad_get_caps() on mad.sink, and it will return audo/mpeg, mpeglayer=[1,3]
[00:09] <Company> for optional fields
[00:09] <dolphy> hmm when using gst_caps_from_string in gst_element_link_filtered the caps has to be freed manually it seems
[00:09] <dolphy> right ?
[00:10] <ds-work> dolphy: yes, because gst_element_link_filtered() takes a const pointer
[00:10] <thomasvs> ds-work: oh, you mean, after I fix mad so that it fixes the layer after negotiation, I can use that function to get the actual layer ?
[00:10] <ds-work> thomasvs: no, you can use that currently to see that mad doesn't know the layer (or doesn't specify it)
[00:11] <dolphy> yup
[00:12] <thomasvs> ds-work: so even though the caps got fixated to being layer 1, that function will still report [1,3] ??
[00:12] <ds-work> thomasvs: yes
[00:12] <thomasvs> ds-work: hm.  how does it remember it just took a guess ?
[00:14] <ds-work> remember what?  get_caps() just calls the getcaps function on the pad and/or returns the pad template caps
[00:16] Action: ds-work wanders off to lunch
[00:22] <Company> thomasvs: i think ds-work didn't get that you want to extract useful user information from the caps
[00:23] <Company> thomasvs: but if you used src ! mp3parse ! mad the second link should have the correct layer set
[00:23] <Company> if not it'd be an mp3parse bug
[00:28] <ds-work> hmmm... decoders might as well set the correct caps on sink pads
[00:28] <ds-work> it adds a burden on the plugin coder, though
[00:29] <thomasvs> ds-work: well, it creates a different problem
[00:30] <thomasvs> ds-work: when do you know you have the right caps ?
[00:30] <thomasvs> ie, you get two times a caps deep notify, the first because of fixate, but wrong, and the second because of the plugin fixing it
[00:30] <thomasvs> ds-work: so, not really sure how I can reliably get the layer
[00:30] <thomasvs> Company: ok, I guess I have to rewrite some things, and for each kind of info I need to extract, I need to use a different pipeline then
[00:41] Nick change: LeRoutier -> teletubies
[00:41] Nick change: teletubies -> Corbier
[00:43] <dolphy> hmm i m having troubles to get EOS
[00:49] markov ([email protected]) left irc: "good nite, folks"
[00:53] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[00:55] sublett ([email protected]) left irc: "I like food, food is good!"
[01:00] <iain> Company: yes, it does do a notify automatically
[01:11] kmaraas ([email protected]) left irc: Read error: 110 (Connection timed out)
[01:12] Nick change: Corbier -> LeRoutier
[01:50] <LeRoutier>  
[01:50] <LeRoutier> got something strange with gst-typefind
[01:51] <LeRoutier> for audio asf (wma) : video/x-ms-asf  || for video+audio asf (wmv) : No type found
[01:52] <LeRoutier> so, asfdemux might have some problem no ?
[01:52] <LeRoutier> argh, no
[01:52] <LeRoutier> forget what i wrote in the past 3 lines
[01:54] <Company> typefinding and actual decoding are 2 completely different subsystems
[01:55] <Company> so there might be types found for stuff we don't decode and there might be types not found for stuff we do decode
[01:55] <Company> the second should be fixed btw ;)
[01:55] Company ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[01:56] <LeRoutier> yep, but calling gst-typefind on a directory instead of the file would always return "No type found" i hope
[01:57] Company ([email protected]) joined #gstreamer.
[02:00] <LeRoutier> <LeRoutier> yep, but calling gst-typefind on a directory instead of the file would always return "No type found" i hope
[02:01] <mathrick> nite
[02:02] Nick change: mathrick -> mathrick|sleep
[02:12] <Company> no clue
[02:12] <Company> maybe it returns application/x-directory
[02:13] <LeRoutier> nope, it returns "No type found"
[02:14] <LeRoutier> i wrote all that because i made a mistake and ran gst-typefind on the directory containing wmv files but not files themselves
[02:16] <Company> yeah
[02:17] <Company> you need to run /path7to/dir/*
[02:18] Zeenix ([email protected]) left irc: "Leaving"
[02:23] walters ([email protected]) joined #gstreamer.
[02:26] Rotty ([email protected]) left irc: Remote closed the connection
[02:29] <LeRoutier> hum. got an mp3 which type is not detected by gst-typefind but for which 'file' finds it beeing a mp3
[02:29] <LeRoutier> mplayer plays it well too
[02:29] <Company> gimme
[02:34] Action: wheels thinks there should be an online repository of b0rked files for stress testing media frameworks.  :-)
[02:35] <LeRoutier> wheels, a porn repository ?
[02:35] <wheels> LeRoutier: Nah, those are usually encoded with great precision.  ;-)
[02:37] <Company> we need to build such a repo
[02:37] <Company> should be reasonably easy
[02:37] <taaz> wheels: just write a GstCorrupter element that flips bits based on some random function
[02:37] <Company> yeah
[02:38] <taaz> though that's certainly not the best way to do things
[02:38] <Company> and put files that crash gst though they shouldn't into that repo
[02:38] iain ([email protected]) got netsplit.
[02:38] <taaz> since you really want to flip specfic flags in media to see what happens
[02:38] <wheels> taaz: You underestimate the creative forces of the internet in their ability to corrupt things.
[02:38] <Company> well, just flip randomly and let it run for a night
[02:38] <Company> or a week
[02:38] <taaz> you can't be too random or it's a lost cause
[02:38] foser ([email protected]) left irc: "[ I want to believe ]"
[02:38] <taaz> ideally you want error recovery
[02:39] <Company> yeah
[02:39] <wheels> (At one point I found that the worst file that screwed up things on TagLib was a 50 mb empty file -- it kept searching looking for either a sync or a tag header...)
[02:39] <taaz> using randomsrc is silly since start codes and such will never occur
[02:39] <wheels> or "blank" rather
[02:39] <taaz> speaking of blanknesss...
[02:39] <wheels> taaz: the tough thing is that you have to get stuff that has the right syncs in some places but is screwed up elsewhere...
[02:40] <taaz> what's the computation cost of detecting a 0 sequence?
[02:41] <taaz> i ask because of the specific case of p2p software that creates a sparse file and fills it with data as it gets it
[02:41] <taaz> it's done at a low level where you can't check via an api, or at least i don't know how to
[02:41] <wheels> taaz: I don't know a reliable way to check other than seeking...
[02:41] <wheels> taaz: which sucks on a large sparse file...
[02:42] <taaz> but ideally a player would see blocks of 0s and just skip them so you could start to watch/listen to whatever is actually in the file
[02:42] <taaz> reading ahead is fine, it's just i'd like to stop the void data from going into decoders
[02:43] <taaz> just a thought.  most players vomit on such data.  totem seems to do well.  xine, strangely, doesn't
[02:43] <wheels> taaz: in that case you could cheat a little bit -- maybe reading every 10k or so...
[02:43] <taaz> gst-player last i checked lost it's mind
[02:43] <wheels> taaz: (since you wouldn't really care about small blocks)
[02:43] <taaz> wheels: how about just matching the bs of the disk...
[02:44] iain ([email protected]) got lost in the net-split.
[02:44] <ds-work> check to see how tar or gzip do it
[02:45] <LeRoutier>  
[02:45] <LeRoutier> Company, found 2 other mp3 that are not detected as such
[02:45] <Company> how about writing a giftsrc or whatever file sharing you use and query info about the file there
[02:45] <Company> LeRoutier: i'm currently looking at that one only :)
[02:45] <taaz> Company: that's too hard
[02:46] <taaz> i already know i just want to skip 0'ed data
[02:46] <ds-work>  /* Sparse file processing.  */
[02:46] <ds-work>   for (counter = 0; counter < BLOCKSIZE; counter++)
[02:46] <ds-work>     if (buffer[counter] != '\0')
[02:46] <ds-work>       return 0;
[02:47] <ds-work> that could be made faster by about a factor of 100
[02:47] <taaz> i would have written an element to do that already but all this caps and seeking and other gst stuff scares me too much ;)
[02:47] Shoragan_ ([email protected]) left irc: "Leaving"
[02:47] <taaz> 100? how so?
[02:48] <ds-work> check a few bytes to see if it's trivially non-zero
[02:48] <ds-work> then use a MMX loop ANDing data together
[02:48] <ds-work> every few cycles, test for 0
[02:49] <taaz> yeah, but the you have an issue on how to optimize the size of inner loops
[02:49] <taaz> like how many iterrations
[02:49] <taaz> probably varies per arch
[02:49] <ds-work> who cares?
[02:49] <ds-work> use a lib for it
[02:49] <wheels> ds-work: your bottleneck isn't going to be on the testing...
[02:49] <wheels> ds-work: and switching into and out of MMX mode is slow...
[02:50] <wheels> ds-work: You can test a bajillion bytes for 0 quickly, what you're worried about is optimizing the number of disk reads / seeks.
[02:50] <Company> stupidly checking 600MB for non-0 is slower
[02:51] <Company> and you'Re not gonna have disk reads on linux
[02:51] <Company> because there's no sectors allocated yet for empty data
[02:51] <Company> it's only slow (and non-NULL, too) on windows
[02:51] <ds-work> on Linux, you could probably peek into the page tables to see if a page is mapped to the 0 page
[02:51] <wheels> Company: hmm, true
[02:52] <wheels> ds-work: but then you get (quickly) into portablility issues
[02:52] <ds-work> wheels: use a lib for it
[02:52] <taaz> ds-work: you keep refering to some magic lib ;)
[02:53] <Company> use liboil!
[02:53] <Company> in the future we'll just call oil_run (gstreamer) in gst_init and be done
[02:54] <taaz> and what will that do?
[02:55] <Company> run an arch optimized gstreamer written in assembler, d'oh
[02:59] <LeRoutier> got gst-typefind complaining about optsched in error state  on 4 mpg file for which it says "No type found"


-------------------------------------------------------
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