IRC Logs

*******************************************************************
[03:15] jcsston ([email protected]) joined #gstreamer.
[03:34] <thaytoo> happy April 1, everyone!
[03:35] <thaytoo> for today only, our LUG is selling SCO Intellectual Property Licenses for Linux with a special $1 discount if anyone wants in ;)
[03:38] <steve_b_> that GStreamer in KDE is an april fool, innit?
[03:44] <ds-work> maybe
[03:45] Action: Company shoots debian's mozilla
[03:45] Action: ds-work shoots mozilla
[03:46] Action: ds-work is about to ask a really dumb question
[03:46] <ds-work> how do I compile khtml separately from the rest of KDE?
[03:50] <Company> developer.apple.com?
[03:50] <ds-work> not a bad idea, actually
[03:56] <Company> gnomes panels suck
[03:56] <Company> putting them somewhere doesn't work 90% of the time
[03:57] <Company> and i hate nautilus saying "galeon" in the titlebar
[03:58] <Company> but i'm obviously an expert and not a usability guy
[03:59] <Company> how can i get rid of gkrellm and the window list?
[03:59] <Company> s/and/in/
[04:00] Action: ds-work #ifdef's all mmap code in gstreamer
[04:02] The_Company ([email protected]) joined #gstreamer.
[04:03] <The_Company> hi, i'm the new debian machine
[04:03] <Company> hi debian machine
[04:04] BBB ([email protected]) joined #gstreamer.
[04:06] The_Company ([email protected]) left irc: Remote closed the connection
[04:07] The_Company ([email protected]) joined #gstreamer.
[04:08] <BBB> ping anyone?
[04:08] <The_Company> pong
[04:08] <Company> pong
[04:08] <BBB> is it fine if I add an #ifdef GST_UNSTABLE_API around not-stable (0.9.x) API in CVS?
[04:09] <BBB> that means it's there for me but not for general use, which is what I want
[04:09] <BBB> I'll make a note in the documentation that it's non-stable API
[04:12] <The_Company> i'm not sure how we handle that api when plugins link to it
[04:12] <The_Company> (static linking?)
[04:12] <The_Company> but other than that its fine with me
[04:14] <ds-work> unstable API implies unstable ABI.  we don't want that
[04:15] <BBB> that's what it means
[04:15] <BBB> "don't use this"
[04:15] <ds-work> but then you use it
[04:15] <BBB> unless you know what you're doing, right?
[04:16] <BBB> I use it in gst-rec, and I'll add a hack to configure.ac to enable it for all local plugins for me
[04:16] <ds-work> and then the user upgrades gstreamer without upgrading gst-rec, and boom!
[04:16] <BBB> and maybe you guys like it and then we can stabilize the API, right?
[04:16] <BBB> gst-rec isn't an end-user product yet :)
[04:17] <BBB> I didn't get comments on the patch in bugzilla, so either people don't look or it's ok
[04:17] md` ([email protected]) joined #gstreamer.
[04:17] <ds-work> which bug?
[04:17] <ds-work> we have so many...
[04:18] <BBB> 118142
[04:22] <BBB> profiles for plugins, profile lib (like the one in gnome-media, but imo nicer), example app to retrieve profiles per plugin, GNOME UI for profile editing
[04:22] <BBB> that's what they do
[04:29] The_Company ([email protected]) left irc: Remote closed the connection
[04:29] Company ([email protected]) left irc: "Client exiting"
[04:31] Company ([email protected]) joined #gstreamer.
[04:32] <Company> eek
[04:32] <Company> "Anwendungen"
[04:36] Company ([email protected]) left irc: Remote closed the connection
[04:37] <ds-work> BBB: how do you handle profiles that require a particular video size?
[04:38] Company ([email protected]) joined #gstreamer.
[04:42] <BBB> I don't think you can enforce that... we discussed that on the mailinglist and nobody had a solution
[04:42] <BBB> I assume you mean VCD, SVCD and so on?
[04:45] <ds-work> yeah
[04:49] <BBB> I don't have a solution yet :/
[04:50] <BBB> I assume that videoscale would implement profiles and offer those options
[04:50] <ds-work> ugh
[04:50] <BBB> yes, I know
[04:50] <ds-work> videoscale doesn't provide properties for size, either
[04:50] <BBB> do you have something better?
[04:50] <ds-work> maybe
[04:50] <BBB> something that's configurable for end users etc.?
[04:50] <BBB> then we extend profiles with filtercaps or so
[04:51] <ds-work> I'm thinking of something along the lines of a bin containing a launch line
[04:51] <ds-work> in concept
[04:51] <ds-work> not necessarily in implementation
[04:51] <ds-work> so a VCD profile might be
[04:51] <BBB> that's what thomas has, but it is unworkable for end users
[04:52] <BBB> in his version, you need to know gst-launch syntax to use it at all
[04:52] <BBB> that's not good
[04:52] <BBB> so it's fine as a concept
[04:52] <BBB> but you need more sensible control, like my UI, for end users
[04:52] <ds-work> videoscale ! video/x-raw-yuv, width=100, height=100 ! mpeg2enc property1=42
[04:53] <BBB> it'd be great to use something like an gst-editor interface to set this up
[04:53] <BBB> but that'd become very difficult, I'm affraid
[04:54] <BBB> anyway, concepts vs. implementation
[04:54] <BBB> I guess
[04:56] <thaytoo> whee! off to Europe in the morning
[04:59] <ds-work> I don't think profiles based on individual element is going to work.
[05:00] <Company> hf thaytoo
[05:00] <BBB> for specific cases, it's difficult
[05:00] <BBB> like VCD
[05:00] <Company> and don't suffer too much from jetlag :)
[05:00] <BBB> however, for something like divx encoding or so, it makes a lot of sense
[05:00] <thaytoo> Company: sure will, and I hope not :)
[05:00] <BBB> 'divx high quality'
[05:01] <BBB> 'divx low quality'
[05:01] <BBB> (just as examples)
[05:01] <BBB> and then 'mp3 high quality'
[05:01] <BBB> 'mp3 CD quality'
[05:01] <BBB> whatever
[05:01] <thaytoo> Company: I'm hoping to meet Martin Soto for a day in Koblenz when we pass through
[05:01] <BBB> if you show one profile to combine all those, it gets very messy
[05:01] <BBB> so in some cases, combinations are good. in other cases, separate element profiles are better and combinations don't make sense
[05:02] <Company> thaytoo: if you do, drop me a mail and i'll show up
[05:02] <thaytoo> cool :)
[05:02] <Company> at least if i have the time
[05:02] <Company> the closer to an airport the better ;)
[05:03] <Company> autogen should check for availability of the "patch" command
[05:03] <Company> ;)
[05:09] <Company> since when does the core need a C++ compiler?
[05:10] <ds-work> Company: it doesn't.  but autoconf tests for it
[05:11] <Company> can we avoid that without much trouble?
[05:11] <ds-work> no, it's part of the LIBTOOL macro
[05:11] <ds-work> along with f77, etc
[05:11] <ds-work> iirc
[05:12] <Company> ah right
[05:12] <Company> that AC_PROG_CC thingy
[05:12] <Company> never rebuild a reiser tree
[05:13] <ds-work> AC_LIBTOOL, I think
[05:13] <ds-work> gstreamer has a lot of dependencies when you're compiling from source
[05:13] Action: ds-work remembers the bad old days
[05:14] <ds-work> compiling stuff for windows is not really fun
[05:14] <ds-work> I really wouldn't mind having apt-get about right now
[05:14] <Company> heh
[05:15] <Company> guess what i'm doing
[05:15] <Company> ./autogen - whoops - ap-get install ... - repeat
[05:15] <Company> gstreamer has a lot of build deps, but very little runtime deps (at least the core)
[05:16] <ds-work> popt, glib, bison, flex, libxml2
[05:16] <ds-work> perl
[05:17] <ds-work> not including doc stuff
[05:17] <Company> autotools, gettext and firends
[05:17] <ds-work> I'm only building dist tarballs
[05:17] <ds-work> i really don't want to get into using autotools on Windows
[05:17] <ds-work> that's just asking for more pain please
[05:18] <Company> hey, we forgot one tool (i don't have installed yet)
[05:18] <Company> make
[05:19] <Company> there's a huge difference between setting up LFS and Debian
[05:19] <Company> after you built your LFS you've got all the build tools
[05:20] <ds-work> install build-essential
[05:22] <Company> important meta-packages should be easier to find (or maybe i should have a look into tasksel, aptitude or so)
[05:23] <Company> what do you think, how hard would it be to get automatic package downloads working for missing codecs in Debian?
[05:26] <Company> so when you try to play an mp3 you get a dialog box "You're trying to play a file of format audio/mpeg but don't have the right plugins for it. Should I attempt to install them for you?"
[05:27] <ds-work> it would be nice to automate the process of 'File 'media' could not be played, please install package X to continue'
[05:27] <ds-work> installing directly is probably bad
[05:28] <ds-work> I talked to taaz about it, but he didn't really get the issue
[05:28] <Company> well, you could offer to install after giving root password when a gconf option is set or so
[05:28] <Company> but you'd need a centreal repository in that case
[05:29] <ds-work> it's still something separate from gstreamer, though
[05:29] <ds-work> it should be a separate app, so that distributors can put the appropriate code in
[05:29] BBB ([email protected]) left irc: "Client exiting"
[05:30] <ds-work> apps could start 'gstreamer-missing-codec' if there
[05:30] <Company> s/app/lib probably
[05:30] <ds-work> 's a missing codec error
[05:30] <Company> or that
[05:30] lilo ([email protected]) left irc: "bbias"
[05:31] <ds-work> also, how do we list all the media types we handle?
[05:31] <dholmes> It would be great if the app could wait while the codec installs and then start playing right away, like in Windows Media Player
[05:31] <ds-work> nautilus needs it
[05:31] <ds-work> dholmes: not until we rewrite the registry code
[05:32] <dholmes> ds-work, sure, just throwing it out there as a long-term goal
[05:32] <Company> yeah, new registry code is a must for that
[05:33] <Company> but then it's just a matter of having a central registry <-> package mapping and writing a wrapper for that
[05:33] lilo ([email protected]) joined #gstreamer.
[05:34] <ds-work> sort of
[05:34] <ds-work> actually, we'd need a media type -> package mapping
[05:35] <ds-work> and ideally, it would be local to the machine
[05:35] <Company> we'd need a plugin/element -> pacakge mapping
[05:35] <ds-work> er, no
[05:35] <Company> that would be easiest at least
[05:35] <ds-work> how does spider know that video/mpeg is handled by mpeg2dec if mpeg2dec isn't installed?
[05:36] <Company> we'd need that mapping external to the registry
[05:36] <Company> you could get everything you need if you once installed all possible plugins and used that registry file
[05:40] <Company> I WANT RHYTHMBOX DAMNIT
[05:40] <Company> (sorry)
[05:44] sub_pop ([email protected]) left irc: "Client exiting"
[05:46] sub_pop ([email protected]) joined #gstreamer.
[05:49] <Company> hrm, evolution is unusable
[05:49] <Company> i want an option like "keep messages on server unless i delete them locally
[06:03] Nick change: tim__ -> bitshifter
[06:11] <dilinger> Company: imap?
[06:12] <dilinger> people who use "keep messages on server" w/ pop3 do horrible things to mail servers.   pop3 wasn't designed for that
[06:13] <Company> dilinger: i'm not sure our uni servers do imap :/
[06:18] <ds> pop3 wasn't designed for anything, it seems
[06:19] <ds> when I admin'ed a mail server 8 years ago, I turned off pop3 and told people to deal
[06:19] <dilinger> it was designed for storing messages for a short period of time
[06:19] <dilinger> i worked on popa3d a bit, it's a clean little protocol
[06:20] <ds> that passes cleartext passwords
[06:20] <dilinger> nah, there's extensions for TLS
[06:20] <ds> that nobody supports
[06:20] <dilinger> evolution does ;)
[06:21] <dilinger> and server side, courier does
[06:21] <dilinger> and that's all i care about, everyone else can use their crappy plaintext-only MUAs
[06:21] <dilinger> ;p
[06:22] jcsston` ([email protected]) joined #gstreamer.
[06:23] Parn ([email protected]) joined #gstreamer.
[06:23] <Parn> hi hi
[06:23] <Parn> a question about totem... is totem in sync with gstreamer 0.8?
[06:23] <Parn> which version? maybe CVS? maybe not at all?
[06:24] ChrisHJW ([email protected]) left irc: Ping timeout: 14400 seconds
[06:26] <Company> Parn: I think totem cvs works with gstreamer 0.8
[06:30] <Company> gah
[06:30] Action: Company shoots his uni
[06:35] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[06:36] ds ([email protected]) left irc: Operation timed out
[06:38] ds ([email protected]) joined #gstreamer.
[06:43] arana ([email protected]) joined #gstreamer.
[06:52] Action: dilinger wonders how taaz's 0.8 packages are going
[06:57] spyder482 ([email protected]) left irc: Nick collision from services.
[06:57] Nick change: arana -> spyder482
[07:03] ChrisHJW ([email protected]) joined #gstreamer.
[07:06] thaytoo ([email protected]) left irc: "later"
[07:50] ser-ver ([email protected]) joined #gstreamer.
[07:52] ser-ver ([email protected]) left #gstreamer ("Leaving").
[08:00] <dholmes> I just wanted to say, gstreamer is really, really cool :D
[08:07] <ds> it goes great with ice cold beer, too
[08:09] Action: bitshifter sometimes wishes google would index bugzilla.gnome.org
[08:11] <ds> Disallow: /
[08:14] <bitshifter> If I could make it show only those products I'm actually interested in on the query page, my quality of life would improve significantly
[08:15] ChrisHJW ([email protected]) left irc: Read error: 110 (Connection timed out)
[08:15] ChriHJW_log ([email protected]) left irc: Read error: 110 (Connection timed out)
[08:15] <ds> bitshifter: you can always copy the source of the query page and remove the bits you don't want
[08:16] <bitshifter> ds: that's a brilliant idea actually, I never thought of that
[08:16] Action: bitshifter tries
[08:17] <ds> you'd want to update it occasionally
[08:17] caro ([email protected]) joined #gstreamer.
[08:17] <ds> I use preset queries for almost everything.  bugzilla's queries are too cumbersome
[08:31] Rotty ([email protected]) joined #gstreamer.
[08:33] <bitshifter> ds: works great, thanks again 
[08:34] spyder482 ([email protected]) left irc: "Leaving"
[08:42] Misirlou ([email protected]) left irc: Read error: 54 (Connection reset by peer)
[08:42] kmaraas ([email protected]) left irc: "Leaving"
[08:49] adar ([email protected]) joined #gstreamer.
[08:49] <adar> this may be the wrong place for the question, but does anyone know why Debian unstable is still on gstreamer 0.6, not 0.8? Is 0.8 coming in only with the rest of gnome 2.6/
[08:50] Misirlou ([email protected]) joined #gstreamer.
[08:55] aeyakovenko ([email protected]) joined #gstreamer.
[08:55] <aeyakovenko> hello
[08:56] <aeyakovenko> how come v4l plugin doesnt work anymore, i cant get this to run v4lsrc ! xvimagesink.  
[08:57] <bitshifter> adar: packaging 0.8 is in process as far as I know (not exactly a trivial task from what I gather)
[09:06] aeyakovenko ([email protected]) left #gstreamer.
[09:22] markey ([email protected]) joined #gstreamer.
[09:31] Nick change: mathrick|sleep -> mathrick
[09:31] <mathrick> re
[09:36] markey ([email protected]) left irc: "leaving"
[09:47] lilo ([email protected]) got netsplit.
[09:50] adar ([email protected]) left irc: "Leaving"
[09:53] lilo ([email protected]) got lost in the net-split.
[09:54] lilo ([email protected]) joined #gstreamer.
[09:59] ChrisHJW_log ([email protected]) joined #gstreamer.
[10:00] jdahlin ([email protected]) left irc: "Leaving"
[10:05] thomasvs ([email protected]) left irc: Read error: 110 (Connection timed out)
[10:13] thomasvs ([email protected]) joined #gstreamer.
[10:25] KoRnouille ([email protected]) joined #gstreamer.
[10:36] jcsston` ([email protected]) left irc: Read error: 60 (Operation timed out)
[10:38] jdahlin ([email protected]) joined #gstreamer.
[10:38] caro ([email protected]) left #gstreamer.
[10:48] dolphy ([email protected]) joined #gstreamer.
[10:48] <dolphy> grmbl gstplay seems broken...
[10:48] <thomasvs> dolphy: ah, so I'm not the only one ?
[10:49] <thomasvs> dolphy: and you finally have time again for your lost sheep too ? :)
[10:51] <dolphy> well time time...
[10:51] <dolphy> i have generated a trace and i need to look at it\
[10:52] <dolphy> have to pay some american budies first :)
[10:52] Nick change: mathrick -> mathrick|Uni
[10:53] <thomasvs> well, I have theora rpms, and I'd like to test theora from the player but I need to get it working first :)
[11:06] sublett ([email protected]) joined #gstreamer.
[11:20] wheels ([email protected]) left irc: "using sirc version 2.211+KSIRC/1.3.9"
[11:37] thaytan ([email protected]) joined #gstreamer.
[11:53] markey ([email protected]) joined #gstreamer.
[11:53] Shoragan ([email protected]) joined #gstreamer.
[12:04] ChrisHJW ([email protected]) joined #gstreamer.
[12:09] ChriHJW ([email protected]) joined #gstreamer.
[12:11] jonathang ([email protected]) joined #gstreamer.
[12:13] <jonathang> thomasvs, when using a pipeline like "gst-launch --gst-scheduler=basicgthread filesrc location=some.mp3 ! mad ! adder ! alsasink", i still get the infamous SIGFPE
[12:15] markey ([email protected]) left irc: "leaving"
[12:15] <thomasvs> and without the scheduler change ?
[12:17] <bitshifter> I think that problem came up before and it was somewhere in gstadder.c, 460-464 because of adder->channels and/or adder->width being 0
[12:17] <jonathang> with opt it works perfect
[12:17] <jonathang> bitshifter, yes that's right, see my mail "gstadder problem"
[12:18] <jonathang> channels AND width are 0
[12:20] <jdahlin> How can I find out if a pipeline is seekable?
[12:20] KoRnouille ([email protected]) left irc: Read error: 110 (Connection timed out)
[12:21] KoRnouille ([email protected]) joined #gstreamer.
[12:24] <thomasvs> jonathang: it probably means adder wasn't able to negotiate somehow.  so it looks like a scheduler bug.
[12:24] Action: thomasvs tries to reproduce
[12:28] <thaytan> thomasvs: that's nice, but do you have to tell us about it? 
[12:28] <thaytan> ;)
[12:32] <thomasvs> thaytan: not any more than you need to tell me about not telling you :)
[12:32] Action: thomasvs throws out the bed thaytan is supposed to sleep in this month
[12:33] <thaytan> hehe :)
[12:33] <thaytan> speaking of,we fly in about 18 hours so I'll be a bit off 'net contact
[12:34] <thomasvs> thaytan: can you send me a mail then with when exactly you arrive and leave, and some way of contacting you ?
[12:34] <thomasvs> just so I know, because it's a busy month at our aptmt
[12:34] <thaytan> will do
[12:36] ensonic ([email protected]) joined #gstreamer.
[12:40] <ensonic> hi hi
[12:41] <thomasvs> thaytan: what airline are you flying ? ie, are you arriving at bcn or grn ?
[12:41] <thaytan> bcn, easyjet
[12:52] <jdahlin> thomasvs: did you have problems with totem?
[12:53] <thomasvs> jdahlin: with player and gst-player
[12:53] <jdahlin> thomasvs: totem seems to work fine now
[12:53] <jdahlin> I'll try gst-player
[12:54] Action: thaytan wonders how he managed to create a file called <S-F1>!
[13:02] <jdahlin> thomasvs: gst-player works fine here too
[13:02] <thomasvs> jdahlin: I know, I'm the only one having this problem atm.
[13:02] <jdahlin> except some avi movies (divx,ac3?) are very slow
[13:35] decadix ([email protected]) joined #gstreamer.
[13:36] <decadix> none of the GStreamer 0.8 examples work on my system...
[13:36] <decadix> (same with 0.6.4)
[13:38] <thomasvs> decadix: be more specific, which one for example ?
[13:38] <decadix> the playing examples (helloworld, queueN, thread)
[13:39] <decadix> helloworld produces 2 second of output then stops, the others produce no output at all
[13:43] <decadix> gst-player works though
[13:47] <harshy> what module installs gstreamer-libs-0.8.pc
[13:48] <decadix> i guess that wanting to program with GStreamer on a FreeBSD system was just a bad idea.
[13:50] <thomasvs> decadix: why ? the player works, so it seems the examples are just a bit outdated, which can happen.
[13:51] <decadix> i don't know how to use the API, then
[13:51] <jonathang> decadix, there is a well-written manual
[13:51] <decadix> (the documentation is equally outdated)
[13:52] <jonathang> for doing what is done in the example, it's sufficient
[13:52] <thomasvs> decadix: I'm not following.  I think the easiest for you if you want to get to learn gst, is to go through the manual, try each example as you go, and the first one that doesn't work, you either try to fix it or ask us questions that help you fix it or ask us to fix it
[13:53] <thomasvs> decadix: I think that's the easiest route to follow.
[13:53] <thomasvs> decadix: since gst-player works for you, you've already seen that gstreamer can in fact work and it's worth investing some time in it if you want to use it for your apps.
[13:55] <decadix> what a lag
[13:55] <decadix> anyway, the first one that doesn't work is helloworld :)
[13:56] <jonathang> what does not work in helloworld?
[13:56] <decadix> sound output stops after 2 seconds
[13:57] <jonathang> have you tried with another scheduler?
[13:57] <jonathang> have you tried playing with gst-debug ?
[13:57] <decadix> nope
[13:58] <decadix> i naively expected helloworld to work out of the box, so that i could proceed with actual programming ;)
[13:58] <thomasvs> decadix: can you give me the link to the online doc that you checked for the example ?
[13:58] <decadix> sure
[13:58] <decadix> http://gstreamer.freedesktop.org/data/doc/gstreamer/stable/manual/html/index.html
[13:59] <decadix> btw, the examples in that manual seem to be the same as the ones found under examples/
[13:59] <thomasvs> decadix: the example is extracted from the docs to examples/manual
[13:59] <decadix> indeed
[14:00] <thomasvs> decadix: what does gst-launch filesrc location=your.mp3 ! mad ! osssink do ?
[14:00] <decadix> same
[14:01] <decadix> one second of sound, 3 seconds of break, one second of sound, then nothing
[14:01] <decadix> (timing varies)
[14:03] Action: thomasvs is trying too
[14:12] Action: thomasvs starts suspecting the oss probe stuff changes has broken everything for him.
[14:23] dholmes ([email protected]) left irc: Read error: 110 (Connection timed out)
[14:24] Company ([email protected]) left irc: Remote closed the connection
[14:26] Company ([email protected]) joined #gstreamer.
[14:31] Rotty ([email protected]) left irc: Read error: 113 (No route to host)
[14:32] Nick change: harshy -> harshyZ
[14:37] Company ([email protected]) left irc: Remote closed the connection
[14:38] Company ([email protected]) joined #gstreamer.
[14:40] mathrick|Uni ([email protected]) left irc: Read error: 54 (Connection reset by peer)
[14:44] <ensonic> is  gst-compprep-0.8 working for anyone (I do get 3x GLib-GObject-CRITICAL, 2x CRITICAL, 1x WARNING)
[14:44] <Company> ensonic: with 0.8 or hEAD?
[14:45] <ensonic> Company, head
[14:45] <Company> ensonic: taaz was working on it and fixing those for the debs
[14:45] sublett ([email protected]) left irc: "Why are all the really good songs short?"
[14:46] <ensonic> Company, looks like it is related to Bug#: 138635
[14:48] wheels ([email protected]) joined #gstreamer.
[14:48] <Company> ensonic: yeah, might be - i don't have ladspa here
[14:53] Action: Company is wondering why it fails in gstladspa.c:361 though
[14:56] Action: ensonic wonders that too, that plugin seems to be quite different ;-)
[14:56] <ensonic> ** (process:21602): WARNING **: 0 sink pads, 2 src pads not yet supported
[14:58] <decadix> must be an april fool
[14:59] <ensonic> Company: that alienwah seems to have broken default params :-(, I'll try to find out who the autor is and bug them to fix it
[15:00] <ensonic> Company: dvdreadsrc has the same problem as well :
[15:00] <ensonic> (process:21625): GLib-GObject-CRITICAL **: file gparamspecs.c: line 1846 (g_param_spec_float): assertion `default_value >= minimum && default_value <= maximum' failed
[15:01] <thaytan> ensonic: strange, I was using that earlier today with no problem
[15:02] <thaytan> what are you doing to trigger it?
[15:02] <Company> ensonic: what glib version?
[15:02] <ensonic> thaytan: running gst-compprep-0.8
[15:02] <thaytan> interesting
[15:03] <ensonic> Company: glib-2.2.3
[15:04] <ensonic> another problem comes from a ladspa plugin ("ladspa-sinCos" -> 0 sink pads, 2 src pads not yet supported)
[15:04] <Company> that doesn't look like it's the problem
[15:06] <Company> the only thing i could imagine would be G_MAXFLOAT being defined wrong on your box
[15:06] <ensonic> Company: its linux now :-(
[15:07] <Company> could you try running gst-launch dvdreadsrc and check with what variables that function is called?
[15:09] <ensonic> Company, mompl
[15:12] <ensonic> Company: :-( I cant run dvdreadsrc as I dont have a dvdrom-drive
[15:13] <ensonic> Company: gst-inspect runs without warning, I run gst-compprep with --gst-fatal-warnings in gdb to check
[15:13] <Company> ensonic: shouldn't matter since it happens on instantiating - before the error is detectedd ;)
[15:13] <Company> gst-inspect runs but gst-compprep doesn'T ?!
[15:13] <Company> they're both supposed to run the same code the same way
[15:14] <ensonic> right gst-insprect shows no error, but gst-compprep brings them
[15:15] <Company> are you sure your gst-compprep is ok?
[15:16] <Company> that error totally confuses me right now...
[15:17] <ensonic> Company, is there a glib equiv for --gst-fatal-warnings?
[15:17] <Company> --g-fatal-warnings
[15:18] <Company> or you can set some env var
[15:18] <ensonic> Company, gst-compprep build today from todays cvs
[15:18] <Company> i dunno if we support --g-fatal-warnings correctly in gst though
[15:18] <ensonic> Company, you don't
[15:19] <Company> --gst-fatal-warnings should do the same though
[15:19] <bitshifter> export G_DEBUG=fatal_warnings
[15:19] <ensonic> bitshifter, thx, thats it
[15:20] Action: bitshifter would prefer it if GStreamer handled command line args the same way gtk_init(&argc, &argv) does
[15:23] <ensonic> gst-inspect-0.8 ladspa-alienwah-{mono|stereo} shows the probelms as well
[15:26] Action: ensonic is totally puzzled, when running kill
[15:27] Action: ensonic is totally puzzled, when running in on the shell dvdnavsrc generates the first error, but in gdb it is a ladspa-plugin
[15:29] <ensonic> there is nothing wrong with that?: paramspec = g_param_spec_float (argname, argname, argname, 0.0, G_MAXFLOAT, 0.0, argperms);
[15:29] markey ([email protected]) joined #gstreamer.
[15:31] <thomasvs> ensonic: you sure you're gdb'ing the correct gst-inspect-0.8 ?
[15:32] bilboed_skool ([email protected]) left irc: Remote closed the connection
[15:32] sublett ([email protected]) joined #gstreamer.
[15:45] <ensonic> thomasvs: gdb gst-compprep-0.8 (it is installed)
[15:56] <markey> hmm I've still got a big mem leak in my gst engine :/
[15:56] <markey> I suspect it's the identity handoff callback
[15:57] <markey> I've tried to unref the GstBuffer inside of the callback, but this yields critical errors from gst
[15:57] <markey> dunno what else could cause such a big leak
[15:57] <bitshifter> how big?
[15:57] <markey> 50 mb in 30 minutes approx.
[15:57] <markey> while playing
[15:58] <bitshifter> oh okay, that's not the same as My Big Leak then
[15:58] <markey> how big?
[15:58] <markey> 1gig in 2 minutes
[15:59] <bitshifter> well, I construct + destroy typefind/spider pipelines and it leaks like up to a couple of MB/file when ID3 tags are involved
[15:59] <bitshifter> but that's something else then from the sound of it
[16:00] <markey> I guess so
[16:01] <markey> it's pretty clear that I must unref the GstBuffer* (the docs say so too), but why do I get an error then
[16:01] <markey> weird
[16:01] <thomasvs> markey: hm, sounds like maybe the refcount of the buffer is decreased when it's handed off.  can you try just for debugging to do a g_free on the pointer ?
[16:01] <thomasvs> markey: seems like a refcounting bug for handoff
[16:01] <markey> thomasvs: gfree on the GstBuffer*?
[16:01] <markey> g_free even
[16:01] <markey> ok let's see
[16:01] <Company> you must not unref the buffer on handoff
[16:02] <Company> the buffer isn't g_freed, it's mempooled btw
[16:02] <markey> unref doesn't work, like I said
[16:02] <Company> it can't work
[16:03] <markey> woops sry
[16:03] <markey> misread
[16:03] <Company> since we don't know how many callbacks are connected
[16:03] <Company> which means we don't know how often we would have to ref the buffer
[16:04] <markey> and BOOM
[16:04] <markey> g_free( buf ) is not a good idea
[16:04] <markey> freezes
[16:04] <Company> yup
[16:04] <Company> like i said: buffers are mempooled
[16:04] ChrisHJW ([email protected]) left irc: Ping timeout: 14400 seconds
[16:04] <markey> what to do?
[16:05] <markey> something is causing this leak
[16:06] <Company> the only option is valgrinding
[16:07] <markey> well the engine is still pretty simple, not that much code
[16:07] <markey> I don't see much that can cause leaks
[16:08] <Company> it doesn't mean it'S your fault
[16:08] <markey> trouble is we can't valgrind amarok currently
[16:08] <markey> freezes on startup 
[16:10] ChriHJW ([email protected]) left irc: Ping timeout: 14400 seconds
[16:18] <thomasvs> so. anyone care to speculate why just seeking in matrix.avi triggers caps renego ?
[16:34] <markey> GstEngine::handoff_cb( GstElement*, GstBuffer* buf, gpointer )
[16:34] <markey> is there anything else I must free?
[16:34] Action: markey never really knows with gst
[16:35] <Company> markey: there's nothing you must free/unref if you didn't alloc/ref it
[16:35] <markey> didn't alloc anything
[16:35] Action: markey is innocent
[16:35] Action: Company kicks wheels
[16:35] Action: markey pokes wheels with a pencil
[16:37] <Company> wheels broke the kio check...
[16:37] <Company> good that i'm that good with KDE now so i can fix it myself
[16:38] <markey> you'll be assimilated, I can see it coming
[16:38] <markey> come join #kde-devel, you know you want it
[16:38] <markey> Kompany =)
[16:43] jimmy_dean ([email protected]) left irc: Connection reset by peer
[16:44] <wheels> Company : hmm?
[16:44] Action: wheels just starts reading back...
[16:45] <Company> wheels: you broke the gst-plugins kio check, but my mad C++ skills fixed it
[16:45] Action: wheels broke the Windows build at work yesterday, so I've been busy with damage control for the last couple of hours...
[16:45] <wheels> Company : how so?
[16:45] Action: wheels heads for the cvs list...
[16:45] jimmy_dean ([email protected]) joined #gstreamer.
[16:46] <Company> wheels: autoconf check, you set the kde include dir to KDE_PREFIX/include, not KDE_PREFIX/include/KDE
[16:46] <Company> KDE_PREFIX/include/kde even
[16:46] <wheels> Company : Right, because that's correct.  ;-)
[16:47] <wheels> Company : It should be $KDEDIR/include
[16:47] <Company> wheels: and apps should #include <kde/header.h> instead of <header.h> ?
[16:47] <wheels> Company : That doesn't mean that the Debian folks haven't broken the canonical structure though...
[16:48] <wheels> Company : No, there shouldn't exist a $KDEDIR/include/kde
[16:48] <Company> ah
[16:48] <wheels> Hmm, that was wrong...
[16:48] <Company> so the kde folks are busy clobbering the include/k* namespace ;)
[16:49] <wheels> Company : Actually -- there are a couple of projects that use that dir, but none of them are used by Gst...  (Or are even in the default KDE install IIRC)
[16:49] <wheels> Company : Well, KDE recommends installing into /opt or similar.
[16:49] <wheels> Company : Where are those headers on Debian?
[16:50] <Company> /usr/include/kde
[16:50] <wheels> Yeah, they're patching things then.
[16:50] <wheels> I don't think it's possible to do that with just configure options...
[16:50] Action: wheels checks
[16:51] Action: Company fetches the old "you should use pkgconfig" sentence
[16:51] <Company> i'm down to 17 plugins not being built
[16:52] smoke ([email protected]) joined #gstreamer.
[16:57] <wheels> Company : Ah, you didn't submit your change yet?
[16:57] <Company> wheels: nope
[16:58] <wheels> Company : Ok, I'll try to come up with a better solution in a couple of minutes...
[16:58] <Company> wheels: since i wanted to talk to you first
[16:58] <wheels> Company : And I'll give you a patch for testing...
[16:59] <wheels> Company : Oh, and reading back -- well I easily read between the lines to tell how enthusiastic you really were; it's not hard to see past the growing pains.  :-)
[17:03] <markey> right on
[17:07] Action: wheels fetches coffee and will then go back for some more m4 fun.
[17:10] <Company> ds-work: liboil autogen misses a $@  when calling configure
[17:10] markey ([email protected]) left irc: "leaving"
[17:11] <Company> ds-work: and what about using autoreconf instead of autogen?
[17:14] md` ([email protected]) left irc: 
[17:17] trow ([email protected]) joined #gstreamer.
[17:19] <thomasvs> yikes, mad is bad at seeking, tries to reset explicit caps all the time
[17:23] Action: ensonic 's high level connection framework starts to work (yipppie)
[17:27] mxpxpod ([email protected]) left irc: "To quit or not to quit.. wait... was that a question?"
[17:28] mxpxpod ([email protected]) joined #gstreamer.
[17:30] ChrisHJW ([email protected]) joined #gstreamer.
[17:31] sxpert ([email protected]) left irc: Read error: 110 (Connection timed out)
[17:31] <ensonic> there is certainly a bug related to recent changes with gparamspecs.c g_param_spec_float, today I get several 
[17:31] <ensonic> (process:22228): GLib-GObject-CRITICAL **: file gparamspecs.c: line 1846 (g_param_spec_float): assertion `default_value >= minimum && default_value <= maximum' failed
[17:32] <ensonic> now even in own sourcen when e.g. using ladspa plugins
[17:32] <thaytan> uninitialised variable?
[17:33] <thaytan> ie, declaring the parameter but not initialised the instance variable in the init function?
[17:34] md` ([email protected]) joined #gstreamer.
[17:34] <ensonic> thaytan: dont know yet, in my example I do not yet acces the elements parameters
[17:35] <ensonic> it not looks like it is directly related to gst-plugins/ext/ladspa
[17:36] The_Company ([email protected]) joined #gstreamer.
[17:36] Rotty ([email protected]) joined #gstreamer.
[17:36] <ensonic> if anyone has ladspa plugins installed, just start the editor and add a source from ladspa, most immeditely crash the editor
[17:36] Company ([email protected]) left irc: Nick collision from services.
[17:36] Nick change: The_Company -> Company
[17:38] <Company> grrr
[17:38] <Company> someone knows the correct mount options for VFAT and umlaute on germena installs?
[17:39] <Company> german even
[17:49] <jonathang> Company, certainly codepage=<something>
[17:49] <jonathang> cp850 maybe
[17:51] <Company> yeah, i'm trying iocharset=iso8859-15,codepage=850
[17:51] <Company> but Grönemeyer is still written with a block instead of an Ö
[17:51] <jonathang> and with codepage=cp850
[17:52] <jonathang> and no iocharset?
[17:52] kmaraas ([email protected]) joined #gstreamer.
[17:52] <ensonic> Company: shit happens
[17:52] <ensonic> ciao
[17:52] <Company> ha
[17:52] <Company> hahahaha
[17:52] ensonic ([email protected]) left #gstreamer ("Leaving").
[17:52] <Company> iocharset=urf8
[17:52] <Company> iocharset=utf8
[17:53] <Company> that was so dead simple i didn't even try it
[17:53] <jonathang> it's a winxp partition?
[17:53] sxpert ([email protected]) joined #gstreamer.
[17:54] BBB ([email protected]) joined #gstreamer.
[17:55] <Company> no, win98
[17:57] <Company> iocharset is probably what format the filenamses are expected in by the Linux kernel
[17:57] spyder482 ([email protected]) joined #gstreamer.
[17:58] <jonathang> oh yeah...
[17:58] <Company> i think they use utf16 in vfat no matter what
[17:58] <jonathang> good to know :)
[17:58] <jonathang> in win98?
[17:58] <jonathang> surprising...
[17:58] Rotty ([email protected]) left irc: Read error: 60 (Operation timed out)
[17:59] <Company> well, i got my umlaute so i'm happy now
[18:00] <wheels> Company : where are your Qt includes?
[18:00] <BBB> wheels: your Qt/KDE idea rocks ;)
[18:00] <wheels> Company : /usr/include/qt3?
[18:00] <BBB> "if we make it in Qt with all GNOME devs, there's no way we cannot take over both the KDE and the GNOME workd"
[18:00] <BBB> :D
[18:01] <wheels> BBB : Simple, but elegant.  :-)
[18:02] <Company> wheels: yeah
[18:03] <thomasvs> ok, so
[18:03] <thomasvs> if I understand correctly, then, when we seek on mad, we just point at "somehwere in the bytestream" and hope mad can recover from the sync error.
[18:03] <thomasvs> problem is, sometimes it recovers wrongly and tries to change sample rate.
[18:03] <thomasvs> which it probably shouldn't do.
[18:04] <thomasvs> anyone knows libmad a bit better than me and is able to say if there are nicer ways of making it resync ?
[18:04] <Company> ... unless it's an mp3 with a changing samplerate
[18:04] <thomasvs> Company: I don't think those are allowed at all.
[18:05] <Company> i'm pretty sure they are allowed technically
[18:05] <Company> but i don't see why mad would resync wrong
[18:05] <thomasvs> well, I added some debugs, it is resyncing wrong
[18:05] <Company> since the header signature is unique in an mp3 stream afaik
[18:06] <thomasvs> and from what I remember, it is in fact forbidden to change the samplerate
[18:08] Kabuto ([email protected]) joined #gstreamer.
[18:08] <Kabuto> Company: You there?
[18:08] <Company> anyway, correct way to resync would be to check that the next header is ok, too
[18:08] <Company> Kabuto: yup
[18:08] <Company> which is what libmad does
[18:09] <Kabuto> Company: Got the audioconvert patch and now get stereo ogg.  Thx  But still have a problem with oggs pausing in Rythmbox.
[18:10] <Kabuto> Company: When I first start a song it seems to buffer and it skips a few times until buffered.  MP3 buffer too but never skip
[18:11] <bitshifter> what about dodgy mp3 files that have some RIFF/WAV stuff in the header? is it possible that there is extra data between frames that is not sync-safe?
[18:12] <Company> bitshifter: files with RIFF headers are rejected by spider ;)
[18:12] <bitshifter> yeah I noticed that
[18:12] <bitshifter> (good thing)
[18:12] <Company> bitshifter: and otherwise there's wavparse to filter the headers
[18:12] <Company> nope, spider should play them
[18:12] <Company> s/spider/a good autoplugger/
[18:13] <BBB> spider plays them
[18:14] <bitshifter> hmm, from what I remember wavparse refuses to process them
[18:14] <thomasvs> Company: well, it looks to me like the bit bucket isn't being emptied after seeks, so maybe we're just handing mad inconsequent buffers
[18:14] <BBB> thomasvs: at the resync, you have to do something similar to typefinding
[18:14] Action: thomasvs thinks of a test to verify that
[18:14] <BBB> so don't sync on the first valid frame
[18:14] <BBB> because that could be random crap
[18:14] <BBB> chance is like 1% or so
[18:14] <BBB> so get a valid sync and see if the next few frames from then on are valid too
[18:14] <BBB> 3 or so
[18:15] <BBB> and *then* take it as a valid resync
[18:15] <BBB> that's what typefinding for mp3 does, too
[18:15] <BBB> that should work
[18:15] <thomasvs> well, then your seek is wrong, no ?
[18:16] <thomasvs> to give some background, I'm trying to fix the bug where if you play matrix.avi and seek around a lot, at some point you get an error.
[18:16] <Company> thomasvs: we should definitely empty the bitbucket on seeks, yewah
[18:16] <thomasvs> which is due to a bunch of explicit_set_caps being quickly done, and for some reason one of them fails
[18:17] <Company> BBB: libmad should handle the resync (and it does that)
[18:17] <BBB> hm, I only know the mp3parse code
[18:17] <BBB> and that one is definately wrong
[18:17] <BBB> for seeking
[18:17] <BBB> it syncs too easily
[18:17] <BBB> thereby giving wrong caps
[18:20] walters ([email protected]) joined #gstreamer.
[18:22] AlexCONRAD ([email protected]) joined #gstreamer.
[18:22] BBB ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[18:23] decadix ([email protected]) left #gstreamer ("i'm gonna use libxine instead").
[18:33] <thomasvs> hm, so why am I only getting this in avi's, and not in plain mp3's
[18:33] <thomasvs> maybe avidemux is doing the seek wrong ?
[18:35] Rotty ([email protected]) joined #gstreamer.
[18:37] KoRnouille ([email protected]) left irc: Read error: 110 (Connection timed out)
[18:57] <Kabuto> Company: you got time for a question?
[18:58] <Company> Kabuto: go ahead
[18:59] walters ([email protected]) left irc: Client Quit
[18:59] walters ([email protected]) joined #gstreamer.
[18:59] <Kabuto> Company: The off skipping seems to be in alsasink.  osssink seems to work fine.  I found an old bug #103658 which sounds like it but was fixed.
[18:59] <Kabuto> Company: off=ogg
[19:01] <Company> are you just using alsasink or do you set a device?
[19:02] <Kabuto> alsasink is what I set up with gconf as default.
[19:02] <Company> could you try alsasink device=hw:0
[19:04] <alley_cat> or try with a different scheduler, opt in 0.8 was totally broken for alsasink for me
[19:05] <Company> yeah, thgat might be a reason, too
[19:06] <Kabuto> what is the gst-launch command for ogg play.
[19:06] <thomasvs> hm, ok, so mad needs to resync too much, even on audio, so chances are that the tempbuffer isn't being dealt with correctly.
[19:06] Action: thomasvs checks some more
[19:06] <Company> gst-launch filesrc location=file.mp3 ! spider ! alsasink ;)
[19:07] <Kabuto> thx.
[19:07] <wheels> Company : that's rather optimistic, no?  :-)
[19:11] <Company> wheels: what's optimistic? the spider pipeline?
[19:15] Uraeus ([email protected]) joined #gstreamer.
[19:16] <alley_cat> Kabuto: the ogg problem seems to be scheduler related, i get the same with opt, --gst-scheduler=basicgthread plays it fine
[19:17] <Uraeus> hello
[19:23] arana ([email protected]) joined #gstreamer.
[19:27] pb_ (~pb@2002:5160:4550:1:240:95ff:fe30:aa84) joined #gstreamer.
[19:31] wheels_ ([email protected]) joined #gstreamer.
[19:31] wheels ([email protected]) left irc: Read error: 60 (Operation timed out)
[19:32] Nick change: wheels_ -> wheels
[19:33] dholmes ([email protected]) joined #gstreamer.
[19:36] <wheels> Company : Ok, I've got something that still works here and I'm pretty sure will fix your issue -- should I just commit it or would you rather try a patch first?
[19:37] spyder482 ([email protected]) left irc: Read error: 113 (No route to host)
[19:39] <Uraeus> wheels: your 1st of April may got a wrong timestamp :)
[19:39] <Uraeus> s/may/mail/ :)
[19:39] <Company> wheels: just commit it
[19:41] <wheels> Uraeus : it was 4 seconds after midnight.  :-)
[19:41] sub_pop ([email protected]) left irc: "Client exiting"
[19:43] <Company> oh
[19:43] <Company> my guadec paper has been accepted
[19:43] <Company> i'll make Martin Soto write it ;)
[19:45] <wheels> Company : done
[19:45] <wheels> (brb)
[19:53] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[19:53] sub_pop ([email protected]) joined #gstreamer.
[19:54] <Company> wgeyup, kio is built, you jumped through enough hoops ;)
[19:54] <Company> wheels: yup, kio is built, you jumped through enough hoops ;)
[19:54] Action: Company should read what he types before pressing return, especially with autocompletion
[19:58] teuf ([email protected]) joined #gstreamer.
[20:05] michel_v ([email protected]) joined #gstreamer.
[20:05] <michel_v> hello
[20:05] <michel_v> I'm trying to build the gnome-vfs gst-plugin, but I eventually run into a file not found error
[20:05] <michel_v> gstgnomevfssrc.c:51:35: libgnomevfs/gnome-vfs.h: No such file or directory
[20:06] <michel_v> which package would that file come from?
[20:06] <teuf> michel_v: gnome-vfs development package
[20:07] <michel_v> hmm, no 'dev' packages on gentoo :)
[20:07] walters ([email protected]) left irc: Client Quit
[20:07] <michel_v> though I got this file: /usr/include/gnome-vfs-2.0/libgnomevfs/gnome-vfs.h
[20:07] <michel_v> is this the one it's searching for, but not finding?
[20:08] mathrick ([email protected]) joined #gstreamer.
[20:08] <teuf> yep
[20:08] <mathrick> hrmpf
[20:08] Action: mathrick is annoyed at latest Debian upgrade
[20:08] <michel_v> the line just before the error reads:  gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../../gst-libs -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -pthread -I/usr/include/gstreamer-0.8 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/libxml2 -DGST_DISABLE_DEPRECATED -Wall -Wno-sign-compare -march=athlon-xp -O2 -pipe -fomit-frame-pointer -MT libgstgnomevfs_la-gstgnomevfs.lo -MD -MP -MF .deps/libgstgnomevfs_la-gstgnomevfs.Tpo -c gstgnom
[20:08] <michel_v> evfs.c  -fPIC -DPIC -o .libs/libgstgnomevfs_la-gstgnomevfs.o
[20:09] <michel_v> I'll just try copying gnome-vfs.h in one of those dirs
[20:09] <teuf> it's missing a -I/usr/include/gnome-vfs-2.0
[20:09] jdahlin ([email protected]) left #gstreamer ("Leaving").
[20:10] <michel_v> ok, I added that path to LD_LIBRARY_PATH, maybe it'll be enough
[20:11] kmaraas ([email protected]) left irc: Read error: 110 (Connection timed out)
[20:12] <mathrick> I wonder why GDM suddenly decided to run ~/.xsession instead of my Gnome session
[20:12] <wheels> Company : good  :-)
[20:12] <michel_v> ah, not enough. hmm
[20:12] <mathrick> it breaks everything :(
[20:13] <teuf> mathrick: yeah I had the same issue :)
[20:13] <teuf> except that it didn't run anything since I didn't have a .xsession ;)
[20:13] <mathrick> teuf: how did you manage to get out of it?
[20:14] <teuf> mathrick: I used to have several scripts in /etc/gdm/Session to run my gnome session
[20:14] <teuf> I copied the gnome 2.6 one to ~/.xsession
[20:14] <mathrick> hmm, ugly, ugly hack :(
[20:14] <teuf> dunno
[20:14] <teuf> it works, so I don't care
[20:15] <mathrick> sure, but GDM is broken, not working as it should
[20:15] Action: mathrick does that gross hack
[20:17] <mathrick> teuf: btw, do you use proxy server with apt maybe?
[20:17] <teuf> nope
[20:18] <teuf> why ?
[20:18] thomasvs ([email protected]) left irc: Read error: 113 (No route to host)
[20:18] <jonathang> mathrick, the sessions are now in /etc/dm/Sessions
[20:19] <mathrick> teuf: well, here proxy works 100x better than direct connection, so I'd like to make use of it
[20:19] <teuf> ah ok
[20:19] Action: teuf doesn't have such issues ;)
[20:19] <mathrick> jonathang: ahh, good to know. But shouldn't it have migrated automagically?
[20:19] <jonathang> yes... but you know... unstable is sometimes... unstable ;)
[20:19] <teuf> mathrick: the apt-get upgrade I did today ran at 250 kB/s :p
[20:20] <jonathang> when someone will apt-get dist-upgrade from woody to testing, it (hopefully) won't do that :)
[20:20] <mathrick> teuf: well, with my proxy I managed to get 350kB/s at top speed, but can't get it to work with apt
[20:21] <mathrick> which annoys me
[20:21] <teuf> isn't it a matter of setting http_proxy ?
[20:21] <jonathang> and btw, the packages (like gnome-session, xfce4-session, ...) will be installing their files in /etc/dm/Sessions now
[20:21] <mathrick> I tried /etc/apt/apt.conf.d/
[20:21] <jonathang> in a freedesktop format
[20:22] <jonathang> that can be shared by kdm and gdm.
[20:22] <mathrick> and set Acquire::http::Proxy, as explained in manpage
[20:22] <jonathang> we're just in the middle of the migration
[20:22] Shoragan ([email protected]) left irc: Read error: 60 (Operation timed out)
[20:22] <mathrick> but it didn't get it
[20:22] <mathrick> jonathang: good to know
[20:22] md` ([email protected]) left irc: 
[20:22] <mathrick> jonathang: can I just copy my old scripts over?
[20:23] <mathrick> jonathang: or is some xdg magic needed to get'em going?
[20:24] <jonathang> no, there are .desktop files
[20:24] <jonathang> in which you refer to your scripts
[20:24] <jonathang> gnome-session should be up to date now
[20:25] <mathrick> so I just need to upgrade it once again?
[20:25] <jonathang> yup
[20:25] <mathrick> will do
[20:26] <teuf> do you know which distro BBB is using ?
[20:26] Kaetzchen ([email protected]) joined #gstreamer.
[20:27] alley_cat ([email protected]) left irc: Nick collision from services.
[20:27] Nick change: Kaetzchen -> alley_cat
[20:27] <mathrick> no clue
[20:27] <mathrick> my guess is FC1, though
[20:27] <teuf> ok
[20:27] <teuf> "gentoo" would have been nice :)
[20:28] <mathrick> why do you need gentooers?
[20:29] <teuf> I don't need a gentooer, but in the last 3 days, 2 gentoo users have reported a bug which is actually a dup that BBB encountered some time ago
[20:29] <teuf> would have been nice to close those as NOTGNOME->GENTOO ;)
[20:29] foser ([email protected]) joined #gstreamer.
[20:29] <michel_v> all that was needed was to set GNOME_VFS_CFLAGS = -I/usr/include/gnome-vfs-2.0 in ext/gnomevfs/Makefile
[20:30] <teuf> hi foser
[20:30] <teuf> foser: aren't you a gentoo packager ?
[20:30] <foser> i am
[20:30] <teuf> ah cool
[20:30] <teuf> and are you rhythmbox gentoo packager ?
[20:31] <foser> also
[20:31] <teuf> cool :)
[20:32] <teuf> which version of rb is currently available in gentoo ?
[20:32] <foser> 0.6.8
[20:32] <teuf> I got 2 crash reports from gentoo users because rhythmbox couldn't find a value for a key present in its schema
[20:32] <teuf> does that tell you anything ?
[20:33] <foser> teuf: it tells we didn't install schemas probably .. that is fixed..
[20:33] <teuf> http://bugzilla.gnome.org/show_bug.cgi?id=138767 for example
[20:33] <foser> teuf: this is only in gentoo 'testing' .. not stable
[20:33] <teuf> foser: ah ? is there a corresponding gentoo bug ?
[20:33] <foser> yep..lemmesee
[20:33] <teuf> that's great if it's fixed
[20:34] Nick change: arana -> spyder482
[20:34] <foser> teuf: yeah.. there mustve been some issue with the make stuff though, cause it just didn't install it by default.. but this is fixed in 0.7 afaik
[20:35] <teuf> foser: there used to have issues with 0.6.6, but this should be fixed in 0.6.8
[20:35] <foser> tuef : http://bugs.gentoo.org/show_bug.cgi?id=45417
[20:35] <foser> teuf: we got the report for .8 .. but our stable (due to xine) is still behind
[20:37] <foser> teuf: so most users shouldn't encounter it
[20:38] walters ([email protected]) joined #gstreamer.
[20:38] <teuf> foser: yeah, but one of the person that had the issue was someone I told to try a newer version of rb when he reported a crash against 0.6.1 :)
[20:38] <teuf> foser: the schema installation doesn't work as well as other packages though from what I gather form this bug report ?
[20:39] <foser> teuf: that is correct.. 
[20:39] <teuf> why isn't it reported on bugzilla.gnome.org ? ;)
[20:41] thomasvs ([email protected]) joined #gstreamer.
[20:44] <foser> teuf: because the workaround was so easy we never really bothered.. we don't have a proper fix either
[20:44] <teuf> foser:  export GCONF_DISABLE_MAKEFILE_SCHEMA_INSTALL="1" isn't working ?
[20:46] <foser> teuf: i don't think so.. i haven't looked at the issue in a while.. i'll check it out when i add 0.6.10 (gnome 2.6 is first though
[20:55] <michel_v> foser: by the way, are you the maintainer of gst-plugins-gnomevfs?
[20:56] jdahlin ([email protected]) joined #gstreamer.
[20:58] <jdahlin> hoho, nice registry bug
[21:00] <foser> michel_v: yeah
[21:01] <michel_v> foser: than I have a bug for you :) http://bugs.gentoo.org/show_bug.cgi?id=46494
[21:01] <michel_v> oh, you just checked it
[21:02] <foser> michel_v: yeah.. we have search functionality .. this 'problem' got reported several times so far
[21:03] <foser> michel_v: so now you can answer it here ?
[21:04] <michel_v> foser: I searched for gnome-vfs and didn't find the bug, that's why I reported :/
[21:04] <mathrick> jonathang: btw, what is status of gnome 2.6 in debian?
[21:04] <foser> next year or .. ?
[21:04] <michel_v> as for BMG, no. I only unmasked gnome 2.6
[21:04] <mathrick> jonathang: is it hitting / has hit experimental probably?
[21:05] <foser> michel_v: yeah hm.. always provide your 'emerge info' and the actual error output, this is not useful
[21:05] <foser> michel_v: never ever used it ?
[21:06] <michel_v> used it before, but upgraded/replaced with packages from the official portage tree
[21:06] <foser> michel_v: thats what i want to know, why is everyone always avoiding the obvious answer
[21:07] dolphy ([email protected]) joined #gstreamer.
[21:07] <foser> michel_v: your bug is duped
[21:08] <michel_v> ok
[21:08] <foser> now maybe ppl understand why i dislike bmg so much.. always cleaning up their trash
[21:12] <Kabuto> alley_cat: So it that a gst-plugin bug?  Do I need to open a bug report?
[21:12] md` ([email protected]) joined #gstreamer.
[21:12] <alley_cat> Kabuto: it's a scheduler problem, not sure if there's already a bug about it
[21:13] <michel_v> ah, glad that it was just a mistake on my part
[21:13] <michel_v> :)
[21:13] <alley_cat> there definitely are bugs about opt and alsasink not liking each other, though in a different context
[21:13] <michel_v> bye guys!
[21:13] michel_v ([email protected]) left #gstreamer.
[21:13] spyder482 ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[21:16] dholmes ([email protected]) left irc: Read error: 110 (Connection timed out)
[21:16] <Kabuto> alley_cat: so that gstreamer's scheduler not the kernels. right?
[21:16] spyder482 ([email protected]) joined #gstreamer.
[21:17] kmaraas ([email protected]) joined #gstreamer.
[21:18] <alley_cat> gstreamers scheduler, yes
[21:19] <alley_cat> won't hurt if you file a bug i guess :)
[21:23] Shoragan ([email protected]) joined #gstreamer.
[21:28] markey ([email protected]) joined #gstreamer.
[21:30] <Company> hm, can i just copy all files from one partition to another and have my windows still work after that?
[21:31] <Kabuto> alley_cat: Thanks for the help.  Entered a buggie report.
[21:31] <mathrick> Company: highly unlikely
[21:32] <Company> i'm thinking about why it shouldn't work if i make sure the partition is bootable and is still the primary partition
[21:32] <mathrick> Company: windoze likes to stick whole slew of various files into bootsectors and such, it breaks easily
[21:32] <Company> i'll try, it's win98 it might just work
[21:32] <mathrick> Company: it's not so easy, and depends on which windoze version is in question
[21:33] <mathrick> hmm, win98 needs at least sys'ing partition after move
[21:33] <Company> it might just work
[21:33] <mathrick> otherwise, it won't find its precious command.com and will barf @ you
[21:35] <jonathang> mathrick, it is partially in experimental, i think
[21:35] <mathrick> jonathang: any ETA available?
[21:35] <ds-work> !
[21:37] <Company> lol
[21:37] <Company> www.mplayerhq.hu is just soo nice to read
[21:38] <jdahlin> indeed
[21:38] <Company> a'rpi has a midlife crisis and they found a security hole
[21:38] <Company> and fixed it exactly $X seconds after being notified :)
[21:40] <jonathang> mathrick, i really don't know (am using sid @home and @work)
[21:40] <mathrick> k
[21:40] <jdahlin> Company: we should try to lure him to work for gstreamer
[21:40] <jonathang> mathrick, and i admit that i'm a kde user :) (i have gnome on my laptop though)
[21:41] <Company> jdahlin: we shouldn't lure anyone, he can come out of his own interest
[21:41] <Company> jdahlin: i'm not even sure his personality fits gstreamer
[21:41] <mathrick> jonathang: ahh, I didn't know... sorry ;P
[21:41] <jonathang> :)
[21:42] <jdahlin> Company: maybe not, but he seemed to be interested towards making a multimedia library
[21:42] <jdahlin> Company: and after all I think he knows quite a bit about multimedia
[21:45] <Company> jdahlin: if he does just write plugins, i'm fine with it - but if he tries adding weird API just to be 0.1% faster, i might get annoyed ;)+
[21:46] <jdahlin> Company: I sometimes feel the same
[21:47] <jdahlin> Company: I'm not sure if it makes sense to have all this complex structures, just to avoid GObjects
[21:47] <Company> jdahlin: complex structures?
[21:48] <Company> jdahlin: you mean GstData?
[21:48] <jdahlin> Company: and caps, structure and a few others iirc
[21:48] <jdahlin> all objects that are not GObjects :-)
[21:49] <Company> uh, GstStructures and GstCaps shouldn't be GObjects
[21:49] <Company> ints aren't GObjects either
[21:50] <Company> though i always wanted to make GstData GObjects
[21:50] <Company> but that would have been political suicide - you remember everyone saying that GObjects are slow?
[21:51] <Company> imagine using something that everyone says is slow as the main data passing mechanism in a media framework with realtime ambitions
[21:52] <jdahlin> no, I've only been around for 2 months
[21:52] <Company> that was general slashdot stuff
[21:52] <jdahlin> I'm in favor of not doing premature optimisations
[21:52] <jdahlin> eg, waiting until someone shows the hard data, profiling etc
[21:53] <Company> same here
[21:53] <Company> as i said, i always wanted gstdata to be a GObject
[21:53] <jdahlin> GObject adds some overhead
[21:53] <Company> and it looks like the objections to that are close to 0 right now
[21:53] <jdahlin> but I don't think it matter in most cases
[21:54] <jdahlin> GstData would make so much sense to have a GObject
[21:54] <jdahlin> and GstBuffer, GstEvent
[21:54] kwm ([email protected]) joined #gstreamer.
[21:56] <Company> yeah, last time we redesigned GstData, there were other maintainers who objected to that
[21:56] <Company> so we ended up with this
[21:57] <Company> wheels: you guys could do me a favour and make your apps shut up
[21:57] <Company> "QPixmap: Cannot create a QPixmap when no GUI is being used"
[21:59] <alley_cat> be happy that's the only stuff you get, i get ten pages worth of warnings about the hicolor icon theme :P
[21:59] <Company> well, that was only one line ;)
[22:00] <Company> it's a bit annoying on gst-register
[22:00] <alley_cat> qt and kde definitely are way too chatty on stdout
[22:01] <mathrick> Gnome 1.x used to be, too
[22:02] <Company> GStreamer 0.6 used to be, too
[22:02] <mathrick> oh, where did the colorful past of 0.6 go to? ;)
[22:02] <mathrick> it was sooo fancy
[22:04] <Company> mathrick: export GST_DEBUG=GST_INIT:5
[22:05] <mathrick> Company: I know, but I don't wanna 2MB log for sinesrc ! fakesink :)
[22:05] <Company> mathrick: that category is just those 2 or 3 lines
[22:05] <Company> mathrick: try it
[22:06] <mathrick> ahh, misread it
[22:06] Action: mathrick tries
[22:06] <mathrick> yay, blast from the past :)
[22:08] jonathang ([email protected]) left irc: "Leaving"
[22:09] <mathrick> hmm, is it possible to break out of more than one level of loop?
[22:09] <mathrick> something like break(2); ?
[22:09] Action: mathrick doubts
[22:09] <Company> mathrick: goto
[22:09] <mathrick> nah, I wanted to avoid it :)
[22:11] <Company> mathrick: 
[22:11] <Company> lvs@lvs:~/cvs/gst-plugins$ find -name "*.c" | xargs grep goto | wc -l
[22:11] <Company> 142
[22:11] <ds-work> break == goto end of loop
[22:11] <Company> half of the gotos are in my code ;)
[22:12] Action: Company adobted Linux kernel programming style
[22:12] <mathrick> I know it's equivalent, but I can pretend it's not :)
[22:12] <mathrick> adobted = adopted by adobe?
[22:12] <ds-work> the linux kernel programming style sucks ass
[22:13] <Company> the linux kernel style does the best cleaning up
[22:13] <mathrick> K&R braces, and 8 spaces tabs... yuck
[22:14] <mathrick> and no typedef'd structs
[22:14] <Company> gnome does all this if (failure) {clean_up; return } else { do_sth; do_more_cleanup; return }
[22:15] <mathrick> and kernel does what? panic("failure"); ?
[22:16] <bitshifter> half of which could probably be eliminated if there was a g_alloca_printf() function
[22:16] <Company> if (failure) {goto err; } else { do_sth; }; cleanup_level1; err: cleanup_level2; return;
[22:18] <mathrick> very asm-like
[22:18] <Company> it avoids duplication of code
[22:18] <Company> so if you alloc one more variable that must be cleaned up, you don't need to check every if
[22:20] Uraeus ([email protected]) left irc: "Client exiting"
[22:27] kmaraas ([email protected]) left irc: Client Quit
[22:30] kmaraas ([email protected]) joined #gstreamer.
[22:32] Nick change: harshyZ -> harshy
[22:41] <Kabuto> K.  I have a /system/gstreamer/0.8/default and a /system/gstreamer/default.  The latter .6?
[22:42] <Kabuto> gconf entries BTW
[23:00] <Company> yeah, the latter is 0.6
[23:01] Action: sxpert compiles gstreamer to see if the "create buffers till out of memory" behavior is still there
[23:02] <sxpert> must be that, as valgring is telling me that it's not leaking :D
[23:03] <Company> that's becuase bbuffers are mempooled
[23:03] <Company> and valgrind can't detect a thing with mempools
[23:04] Action: ds-work suggests running gst-launch with -t
[23:04] <Company> are buffers alloc traced?
[23:04] <Company> even when, that doesn'T solve the problem
[23:04] <Company> (an it's -T anyway ;))
[23:05] <Company> you want to know where the buffer leaked
[23:05] <ds-work> uh, that's hard
[23:06] jcsston ([email protected]) joined #gstreamer.
[23:06] <ds-work> since buffers are passed around quite a bit
[23:06] <sxpert> all I know is that playing oggs eventually fills 256M after 4/5 days
[23:06] <Company> you can trace it a lot easier if you know the buffer was created by filesrc
[23:07] smoke ([email protected]) got netsplit.
[23:08] Action: ds-work suggests using a subclass of dory to watch buffers passing in/out of an element
[23:09] wheels ([email protected]) left irc: "switching clients"
[23:10] wheels ([email protected]) joined #gstreamer.
[23:13] smoke ([email protected]) got lost in the net-split.
[23:22] mcv ([email protected]) joined #gstreamer.
[23:22] <mcv> hi folks
[23:23] <mcv> i've this problem: i recorded on tape a lesson, but there's too noise. can you give me some suggestions on how to do noise redctiuion?
[23:25] <sxpert> mcv: there's a "noise removal filter" in audacity. select a silent noisy bit of file, then launch plugin... should do a fairly good job
[23:33] teuf ([email protected]) left irc: "Client exiting"
[23:36] <wheels> mcv: And note that most "noise reduction" that's used in commercial stuff is a two way system -- it has to be turned on on the way in too...
[23:37] <sxpert> wheels: the one in audacity "learns" what the average noise looks like...
[23:38] <mcv> i'm trying with audacity, but i can't tune the way it does noise reduction. i get too mechanical sounds
[23:39] <sxpert> mcv: yeah, that's normal
[23:39] <sxpert> mcv: you should select just a bit of the sound where "silence" is supposed to be.
[23:40] <sxpert> you'll have to make with what remains... 
[23:40] <sxpert> there's no miracle (that's why studios are supposedly *very* quiet ;D
[23:40] <mcv> hmm, i'm gonna try to learn how to make it with cool edit
[23:41] <sxpert> not sure cooledit can do better...
[23:41] <mcv> it offers a graph to control the process
[23:41] <mcv> me too
[23:41] <sxpert> after all, garbage in, garbage out :D
[23:41] mxpxpod ([email protected]) left irc: "To quit or not to quit.. wait... was that a question?"
[23:42] mxpxpod ([email protected]) joined #gstreamer.
[23:47] <mcv> is it possible to reduce mech sounds?
[23:47] kwm ([email protected]) left irc: "Leaving"
[23:47] <sxpert> http://www.thinkgeek.com/stuff/41/habicase.shtml
[23:48] <sxpert> mcv: the audacity filter is the best there is that I know of at this time
[23:55] <mathrick> wheels: C++ question - how do I realloc() with new?
[23:55] <mathrick> wheels: or is it something else I should be using?
[23:58] <wheels> mathrick: Well, what's the situation?
[23:59] <mathrick> wheels: I need to read in int[][], with no ahead knowledge of elements number
[23:59] <mathrick> wheels: so, sooner or later, I might need to realloc mem
[23:59] <wheels> mathrick: Well, you probably shouldn't be using int[][] in C++.  :-)
[23:59] <wheels> mathrick: std::vector< std::vector<int> >  
[00:00] --- Fri Apr  2 2004
[00:00] <mathrick> wheels: hrmpf, it's not eactly the easiest solution, to migrate it to STL :)
[00:00] <mathrick> wheels: anyway, there should be some way to provide realloc() functionality, right?
[00:00] <wheels> mathrick: not really
[00:00] <markey> vector.resize() works nicely
[00:01] <wheels> mathrick: You're doing things the C way in C++, so there's not really a C++ way to do the C thing.
[00:01] <mathrick> ekhm, C++ is evil, did I say this already? ;)
[00:01] <markey> C++ is a tad complex, but once you master it it's very powerful
[00:01] <markey> really
[00:01] <wheels> mathrick: But funtionally those two should (the vector and the two dimensional array) should have almost the same semantics.
[00:01] <markey> in the beginning it's much cursing, tho
[00:01] <mathrick> wheels: I don't believe noone ever needed to realloc some mem, just in vector implementation for example
[00:02] <markey> the speed of C with the elegance of OOP :)
[00:02] <wheels> mathrick: Well, you just resize or use clear -- arrays are discouraged in C++.
[00:02] <mathrick> but how do you implement resize?
[00:02] <wheels> mathrick: You don't care.  :-)
[00:02] <markey> vector::resize( foo )
[00:02] <mathrick> if you cannot realloc mem, that becomes harder
[00:03] <wheels> mathrick: Well, it may use [re/m]alloc internally.  But that's an implementation detail that the developer shouldn't worry about.
[00:03] <mathrick> wheels: I understand there's std::magic somewhere available for STL implementors? ;)
[00:04] <mathrick> wheels: well, yes, but then I cannot mix it freely with new / delete
[00:04] <mathrick> which sucks
[00:05] <mathrick> markey: that's how I invoke it, I asked about actual implementation :)
[00:05] <wheels> mathrick: Well, again -- you're still thinking like a C programmer.
[00:05] <markey> mathrick: what wheels said
[00:05] <markey> we need to brainwash him
[00:05] <markey> clearly
[00:05] <mathrick> wheels: probably, but I just can't believe realloc is _never_ needed
[00:05] yippi ([email protected]) joined #gstreamer.
[00:06] <mathrick> markey: no way, days when I believed in C++ are over ;P
[00:06] <markey> mathrick: have you never used java, python or something else OOP?
[00:06] <wheels> markey: Well, really -- it takes a while to "unlearn" the C way of things (I was doing C for about 5 years before I started with C++)
[00:06] <mathrick> markey: sure
[00:06] <sxpert> there's nothing that needs C++ that can't be done in straight C... hence, no point in messing with C++ after all :D
[00:06] <mathrick> markey: but python hides all mem issues
[00:06] <markey> right
[00:06] <mathrick> markey: so it's not reasonable to ask about it
[00:07] <mathrick> markey: C++, OTOH, exposes you to it
[00:07] <markey> of course not
[00:07] <mathrick> so, C++ needs some functionality Python does not
[00:08] <mathrick> same with Java
[00:08] Action: Company wonders how you should implement GstBuffer in different languages
[00:08] <sxpert> Company: with "Handles" ?
[00:09] <Company> nah, if you had coded gst in that language
[00:09] <markey> what's special about it?
[00:09] <sxpert> Company: mlt uses a nice buffer infrastructure...
[00:09] <mathrick> in Python, you just implement some magic data type, and that's all
[00:09] <sxpert> dunno how we do it...
[00:09] <mathrick> implement in C, of course :)
[00:09] <sxpert> mathrick: lolo
[00:10] <Company> GStBuffer is just a length and a char array of that size basically
[00:10] <markey> what I loathe about C coding is the constant need for casting stuff
[00:10] <markey> that's a bit PITA
[00:10] <Company> yeah, that sucks
[00:10] <sxpert> Company: how are buffers allocated ? with a power of 2 scheme ?
[00:10] <markey> and the ridiculously long function names, since you must encode the type into the name
[00:11] <Company> that's why people make macros for everything
[00:11] <Company> sxpert: up to the allocator
[00:11] <sxpert> Company: ah, I see...
[00:11] <Company> markey: i like those function names
[00:11] <sxpert> Company: should we use a standard allocator thingie ?
[00:12] <Company> markey: debug_printf ("hi") is nicer than x << "Hi"
[00:12] <Company> sxpert: we use mmap and stuff for buffers
[00:12] <wheels> mathrick: Oh, and just for you I'm reading the STL implementation now.  ;-)
[00:12] <markey> gst_element_do_something  <-- the gst_element part is really not needed with OOP
[00:12] <markey> that's a bit annoying
[00:12] <markey> like hungarian notation
[00:12] <mathrick> wheels: I'm grateful :)
[00:13] <mcv> thanks guys, nite all
[00:13] <sxpert> markey: you end up with gst_element::do_something <-- not much better :D
[00:13] <markey> element.do_something() is crisp :)
[00:13] <mathrick> markey: well, not really. Hungarian notation is brain dead ;)
[00:13] <sxpert> Company: how would you explain the buffer leaks that are not leaks but still are leaks ?
[00:14] <mathrick> sxpert: gamma rays
[00:14] <Company> sxpert: that has to do with the struct holding the pointer to the data (the GstBuffer struct) is mempool allocated
[00:14] <sxpert> Company: and ?
[00:14] <sxpert> Company: it doesn't put it back when it's not needed any more ?
[00:15] mcv ([email protected]) left irc: "Frobnicating"
[00:15] <Company> sxpert: yeah, it just gets rid of the pointer to the buffer
[00:15] <Company> the GstBuffer
[00:15] <sxpert> Company: aaaahhh... heresy !!!
[00:15] <Company> but since that GstBuffer is part of a mempool, the mempool still has a pointer to that area
[00:16] <Company> so valgrind doesn't find it
[00:16] <ds-work> yet another good reason for GObjectifying GstBuffer
[00:17] <markey> is identity possibly leaking mem internally?
[00:17] <Company> ds-work: i'd prefer being able to detect running inside valgrind and disabling mempools in that case
[00:18] <Company> ds-work: i'd certainly overwrite the constructor for GstBuffer to allocate from a mempool
[00:18] <thomasvs> MPlayer 1.0pre3try2
[00:18] <thomasvs> please
[00:18] <thomasvs> why not mplayer 1.0pre3try2welovealphacharsinourreleaseversion
[00:18] <sxpert> thomasvs: ROTFL
[00:19] <jdahlin> thomasvs: someone mailed my about my patches against buildbot :)
[00:19] iain ([email protected]) joined #gstreamer.
[00:19] <sxpert> thomasvs: so, someone broke your nice new car (yet another one - I thought you bought a new one last year ?)
[00:19] <iain> yo
[00:19] Action: ds-work sees that one could change GstBuffer from mem_chunk to malloc with only 3 lines
[00:20] <thomasvs> sxpert: you're probably confusing me with damien
[00:20] <thomasvs> jdahlin: and, good comments ?
[00:20] <sxpert> ds-work: thing is, we really need to re-use those buffers if they are already allocated
[00:20] <jdahlin> thomasvs: nah, he just asked me how I've done a couple of things
[00:20] <jdahlin> thomasvs: and I said "custom patches not accepted upstream"
[00:21] <jdahlin> thomasvs: too bad buildbot isn't really maintained
[00:21] <ds-work> sxpert: that is unrelated
[00:21] <sxpert> thomasvs: uh, ah right :D
[00:21] <sxpert> ds-work: ah, well...
[00:22] BBB ([email protected]) joined #gstreamer.
[00:22] <sxpert> ds-work: the idea here is to see the app keep the same average memory useage, not growing like crazy for some reason
[00:22] <Company> ds-work: changing that with only 3 lines is because of a recent commit of me
[00:23] <Company> ds-work: i changed the gstbuffer.c code because i did exactly that when tracing buffer leaks ;)
[00:23] <ds-work> why isn't there a gst_buffer_alloc() static function?
[00:23] Action: ds-work adds
[00:24] <Company> gst_buffer_alloc?
[00:24] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[00:24] <Company> ah, understood
[00:24] <Company> is the valgrind guy on IRC?
[00:25] <iain> hey Company 
[00:25] Action: BBB is tired
[00:25] <BBB> 3 1/2 hours of talks
[00:25] <BBB> oh my god
[00:25] <sxpert> BBB: lol
[00:25] <Company> hey iain
[00:25] <iain> I made that change you said (adding a separate thread for the sink and adding queues)
[00:25] <iain> but now I'm getting 
[00:25] Action: BBB needs a vacation after today
[00:25] <iain> ERROR    audioconvert( 9517) gstaudioconvert.c(245):gst_audio_convert_chain:<audioconvert> Sink pad not negotiated before chain function
[00:27] <Company> iain: ugh
[00:29] <Company> ds-work: is there a nice debugging category to monitor caps nego?
[00:29] <ds-work> GST_PADS
[00:31] <iain> http://www.abisource.com/~dom/abi-mrt-helper.png
[00:31] <iain> okay, that rules
[00:31] <sxpert> heheh
[00:31] <sxpert> iain: can you say 2004-04-01 ?
[00:32] <iain> I know
[00:32] <iain> but the image rules
[00:32] <sxpert> iain: yeah :D
[00:32] <sxpert> iain: of course, there's that for your hamster :D http://www.thinkgeek.com/stuff/41/habicase.shtml
[00:33] Action: iain doesn't go to thinkshite
[00:33] <Company> iain: i wonder why there doesn't happen any caps nego, i guess you gotta debug it yourself or beg me a _long_ while to look at it ;)
[00:33] <sxpert> iain: you'll be missing on a great idea ;D
[00:34] <iain> Company: will you look at it?
[00:34] Last message repeated 2 time(s).
[00:34] <iain> x100000000
[00:34] <sxpert> lol
[00:35] <ds-work> Company: will you look at it?
[00:35] <iain> Company: can we go to itchy and scratchy land?
[00:35] <sxpert> ROTFL :D
[00:36] <iain> first step...work out how to only show GST_PADS in the debug log
[00:36] <ds-work> --gst-debug=GST_PADS:5
[00:37] <iain> I knew if I begged someone would do it for me
[00:37] <iain> step 2...work out what it means
[00:37] Action: iain looks around
[00:38] <Company> iain: http://www.bash.org/?search=titanic - chose one
[00:38] <sxpert> lol
[00:38] markey ([email protected]) left irc: "leaving"
[00:42] <Company> ds-work: i've got another check for gst_lint
[00:43] <ds-work> soot
[00:43] <ds-work> shoot
[00:43] <Company> ds-work: simple version: check for function calls inside g_assert_* and g_return_*
[00:43] <Company> ds-work: harder version: check they're definitely not an lvalue
[00:44] <Company> s/lvalue/on the left site of a comparison/
[00:44] <Company> i want to get GStreamer working with all this checks disabled
[00:44] <ds-work> does it not work?
[00:44] <Company> currently this doesn't work because people do g_assert (important_function () == 0);
[00:45] <ds-work> really?  I find that incredibly stupid
[00:45] <ds-work> besides, g_assert() should be disabled by '#define g_assert(a) a'
[00:45] <Company> no, it shouldn't
[00:45] <ds-work> let the compiler figure it out
[00:46] <Company> hm
[00:46] <Company> sound like an argument
[00:46] <ds-work> although anyone that writes code that relies on that behavior should be taken out back and shot
[00:47] <Company> #define g_assert(expr)           G_STMT_START{ (void)0; }G_STMT_END
[00:47] abo_ ([email protected]) joined #gstreamer.
[00:50] LeRoutier ([email protected]) joined #gstreamer.
[00:50] <LeRoutier> hello
[00:54] AlexCONRAD ([email protected]) got netsplit.
[00:54] <iain> DEBUG        GST_PADS(11951) gstpad.c(1258):gst_pad_link_negotiate: negotiating link from pad queue.src_1:src to pad marlin-sample-sink:sink1
[00:54] <iain> DEBUG        GST_PADS(11951) gstpad.c(1109):gst_pad_link_ready_for_negotiation: parent queue.src_1 of pad queue.src_1:src is not READY
[00:54] <iain> would that be why its not working? queue.src_1 isn't READY?
[00:55] <sxpert> LeRoutier: hello
[00:56] <wheels> mathrick: Oh, from much earlier (I got distracted) -- from what I can tell about the mess of code that is GCC, std::vector does just use malloc and friends, but allocation is templatized so that different allocation strategies can be dropped in.
[00:56] sublett ([email protected]) left irc: "Why are all the really good songs short?"
[00:56] <Company> iain: yeah
[00:56] Kabuto ([email protected]) left irc: "I like core dumps"
[00:57] <iain> Company: and how would I make queue.src_1 READY?
[00:57] AlexCONRAD ([email protected]) returned to #gstreamer.
[00:58] <Company> iain: set its state to PLAYING i guess
[00:58] <Company> iain: though i guess you did that
[00:59] <mathrick> wheels: hmm, ic. I still think C++ is evil :)
[00:59] <wheels> mathrick: Well, acutually the idea of interchangable allocation strategies is pretty neat.
[00:59] <sxpert> mathrick: same here :D
[00:59] Zeenix ([email protected]) joined #gstreamer.
[01:00] <Zeenix> hi
[01:00] <iain> Company: its been added to a pipeline thats already at the PLAYING state
[01:00] <sxpert> <flamebait>C++ is evil</flamebait>
[01:00] <iain> Company: so I'll need to add element_sync_state?
[01:00] <wheels> mathrick: For instance switching to zone allocation when desired without touching the other code.
[01:00] <Company> iain: yeah
[01:00] <Company> wheels: do you know if an app can detect being inside valgrind?
[01:01] <mathrick> wheels: but thanks for your effort
[01:01] <mathrick> wheels: sure, it has it's good sides
[01:01] <mathrick> wheels: but the thing is, C++ is so huge and elaborate, that amounts of elaborated pitfalls in it is just incredible :)
[01:01] <wheels> Company: with or without dirty hackery?
[01:02] <wheels> mathrick: Well, sure -- it's harder to learn, but then development tends to go faster.
[01:02] <Company> wheels: preferrably without, but if that's not possible...
[01:03] <sxpert> wheels: I have found it to tend to have highly unpredictable behavior at times (from a 400Kb of sources project)
[01:03] <wheels> sxpert: Sure, it's highly unpredictable for C programmers.  ;-)
[01:03] <sxpert> wheels: tsst
[01:03] <wheels> sxpert: I find Fourth to be highly unpredictable.  :-)
[01:03] <mathrick> wheels: I still prefer combo approach - C for low level and time crtical stuff, something nice like Python for rapid developing
[01:04] <wheels> mathrick: Well, but C isn't faster than C++ implicitly.
[01:04] <sxpert> wheels: having learned programming on an HP48, forth is like my mother tongue :D
[01:04] <mathrick> wheels: never said that
[01:04] <sxpert> (of course, I was doing basic at 5 years old, like everybody else :D)
[01:04] Action: mathrick still has to learn forth
[01:04] <thaytan> you're all nutbags. We'll be coding C# by the end of the year. GET USED TO IT!
[01:04] <thaytan> brouhahahahaha
[01:05] <sxpert> thaytan: *you* are the nutbag ;D
[01:05] Action: mathrick waits for Lisp#
[01:05] <mathrick> then I can even switch to .NET ;)
[01:05] <sxpert> mathrick: in a couple centuries ?
[01:06] <Company> thaytan: i intend to do the next gui app i do in either C# or Python
[01:06] <mathrick> sxpert: who knows, .NET seems pretty hot atm, so ppl are trying various crazy ideas
[01:07] <wheels> mathrick: I'm just waiting for it to go out of style before it ever gets adopted.  :-)
[01:07] <sxpert> mathrick: it will fall flat on it's back in a couple years...
[01:08] <mathrick> who knows, MS is pushing it hard
[01:08] <wheels> C# isn't conceptually a bad thing, but I don't think that it's stronger than Java and I don't think it will hit some of the same roadblocks...
[01:08] Action: wheels thinks that at some point byte-compiled languages will catch on, but that we're probably about 8-10 years from that.
[01:09] <jdahlin> wheels: nah, it won't take that long
[01:09] <jdahlin> wheels: java is one of the dominant languages today
[01:09] Action: mathrick notes Lisp is about 50 years old now ;)
[01:09] <wheels> jdahlin: I should have been more specific -- I mean in replacing the traditional C / C++ role...
[01:09] <sxpert> wheels: C# and friends will hit the same roadblock when they'll try "µCLR" processors :D
[01:09] <mathrick> wheels: and it's been byte-compiled since forever :)
[01:10] <wheels> mathrick: Well, I don't think there's a requirement of such for list -- isn't a lot of it interpreted?
[01:10] <wheels> s/list/lisp/ -- reflex
[01:10] <mathrick> wheels: surely it is, but usually it ends up byte-compiled anyway
[01:10] <sxpert> wheels: not a chance, you won't be able to program an OS in C#/CLR
[01:10] <wheels> sxpert: You can get pretty close.
[01:11] <sxpert> wheels: that tiny bit of assembler called the JIT is still a problem...
[01:11] <Company> as long as my C code is more portable than CLR code, I don't care
[01:11] <mathrick> it's like Smalltalk - compilation is incremental
[01:11] <sxpert> wheels: you still need the VM...
[01:11] <jdahlin> wheels: it probably never will
[01:11] <wheels> sxpert: there's no reason that you can't write a JIT in the byte compiled language.
[01:11] <mathrick> sxpert: you won't code OS in C either :P
[01:12] <iain> ** ERROR **: file entryscheduler.c: line 590 (gst_entry_scheduler_chain_handler): assertion failed: (priv->bufpen == ((void *)0))
[01:12] <iain> aborting...
[01:12] <iain> yey
[01:12] <sxpert> wheels: Intel tried something like this with the IAPX-432, which was a major failure...
[01:12] <iain> so many threads, so much fun
[01:12] <wheels> sxpert: Sure.  I don't think the technology is ready for such yet.
[01:12] <wheels> sxpert: But I don't see anything that will stop it from getting there.
[01:13] <sxpert> wheels: security is one of the issues preventing you from doing such a JIT. you'd need a pretty good way to prevent malware from self-compiling itself.
[01:13] <wheels> sxpert: Let's not go into all of the security issues that are solved by moving as many things away from C as possible...  ;-)
[01:14] <tromey> JikesRVM is already a self-compiling JIT.  The technology isn't really that complicated
[01:14] <sxpert> wheels: the ones that are pretty close from it are transmeta. they could probably do a code morphing thing that would go from CLR to whatever microcode structure they use
[01:15] aldug ([email protected]) joined #gstreamer.
[01:16] <aldug> has anyone gotten totem to work with gst-0.8 yet?
[01:16] <iain> I have
[01:18] <aldug> iain: what did you have to hack to get it to compile?
[01:18] <iain> nothing to my knowledge?
[01:18] <alley_cat> use the cvs, there's no totem release that compiles with 0.8
[01:18] <iain> CVS HEAD totem and cvs HEAD gstreamer
[01:19] <aldug> it doesn't like baconvideowidget_marshal_VOID__INT64_INT64_FLOAT
[01:19] <aldug> I'm using gst-0.8 with cvs head totem
[01:19] <mathrick> g'night
[01:19] <jdahlin> aldug: cvs up
[01:19] <iain> that sounds like a totem problem
[01:19] <LeRoutier> aldug, platform ?
[01:19] <iain> nothing to do with gstreamer
[01:19] <BBB> bbl, going home
[01:19] BBB ([email protected]) left #gstreamer.
[01:19] <jdahlin> aldug: I fixed that earlier today
[01:19] <aldug> jdahlin: oh, must have missed that :)
[01:19] Nick change: mathrick -> mathrick|sleep
[01:19] <jdahlin> aldug: using anoncvs?
[01:20] <aldug> nope
[01:20] <aldug> I got the change
[01:20] <jdahlin> cool
[01:20] <aldug> I'm about to try out the ffmpeg stuff
[01:21] <jdahlin> works for me here
[01:21] <jdahlin> I can play avis and stuff
[01:22] <thaytan> jdahlin: and DVDs ;)
[01:22] <aldug> jdahlin: I've gotta try out our python bindings too
[01:23] <aldug> s/our/your/
[01:24] <jdahlin> aldug: cool, 0.7.91 should work :)
[01:24] <jdahlin> I need to spend some time integrating them with pygtk though
[01:25] <LeRoutier> anyone has a clue why media with "uncompressed PCM" audio won't play at all under any gstreamer player ?
[01:26] <thaytan> LeRoutier: too complex a format? ;)
[01:26] <LeRoutier> well, i thought PCM was one of the simplest :)
[01:26] <thaytan> it is the simplest :)
[01:26] <thaytan> (was a joke)
[01:26] <LeRoutier> yep, i know
[01:27] <LeRoutier> all videos i have that had this audio format crashed the demuxer (being avidemix or qtdemux)
[01:27] <thaytan> excellent, comedy guages are synchronised. And now.... on with the show!
[01:28] <LeRoutier> thaytan, aren't you one of the main programmer of gst-editor ?
[01:29] <thaytan> I'm the maintainer by default :)
[01:29] <LeRoutier> could you apply this trivial patch please : http://bugzilla.gnome.org/show_bug.cgi?id=138623
[01:30] <LeRoutier> (without this one, gst-editor elements won't show in gnome menu)
[01:32] <ds-work> LeRoutier: it would help if someone uploaded a relevant movie
[01:32] <LeRoutier> ds-work : for the PCM problem ? i uploaded about 10 of them
[01:34] Zeenix ([email protected]) left irc: "Leaving"
[01:37] <thaytan> ds-work: how can I get bugzilla power to close that bug now?
[01:38] <LeRoutier> thaytan, i can do it as i opened it
[01:38] <LeRoutier> saw you applied it, thx
[01:39] <ds-work> thaytan: what is your bugzilla login?
[01:41] <thaytan> [email protected]
[01:41] <thaytan> LeRoutier: please do
[01:41] <LeRoutier> thaytan, i just closed it. 
[01:42] <thaytan> ta 
[01:42] <LeRoutier> but it should be good if you had 'admin' access for gst-editor module
[01:45] <ds-work> thaytan: you should have The Power now
[01:45] <thaytan> ds-work: sweet, thanks
[01:50] <LeRoutier> got another trivial patch but for gst-recorder. i guess i'll have to wait for ronald to be back
[01:52] <Company> someone should get LeRoutier cvs access - i like people that automatically fix boring things like .desktop files
[01:53] <LeRoutier> well, as long as i don't break the code itself (my C is crappy, that's why i prefer to touch to things like m4/desktop/i18n...)
[01:54] <LeRoutier> does anyone work activelly on textoverlay ? (subs)
[01:54] <thaytan> we can always revert stuff :)
[01:55] <ds-work> LeRoutier: not actively, no
[01:55] <thaytan> I committed mpeg2subt last night for DVD subs
[01:55] <thaytan> but not textoverlay, no
[01:56] <LeRoutier> oh, for dvd sub selection ? already integrated into gst-player/totem ?
[01:56] iain ([email protected]) left irc: "I e you all"
[01:57] <alley_cat> i've been thinking a lot about text subs, first plan was to add ssa to mplayer, but i gave up on that
[01:58] <alley_cat> if no one takes it up for gst soon i'll probably start it
[01:58] <LeRoutier> there are external text subs (ssa, smi, srt, ...) but there are also subs embeded inside avis as a third stream
[01:59] <alley_cat> don't forget ogm and matroska :)
[01:59] <alley_cat> although ogm isn't supported by gst anyway currently
[01:59] <LeRoutier> oggdemux problem ?
[02:00] <alley_cat> "no spec" problem i believe :P
[02:00] <LeRoutier> a too common syndrome
[02:09] <LeRoutier> alley_cat, from this bug report, it says mplayer already supports ssa subs : http://bugzilla.gnome.org/show_bug.cgi?id=100931
[02:13] Nick change: mxpxpod -> mx|gone
[02:15] pb_ (~pb@2002:5160:4550:1:240:95ff:fe30:aa84) left irc: "Client exiting"
[02:15] foser ([email protected]) left irc: "[ I want to believe ]"
[02:15] walters ([email protected]) left irc: "out"
[02:23] <thaytan> LeRoutier: not yet integrated into gst-player and totem
[02:24] <thaytan> there's a gst-launch pipeline in ext/dvdnav/gst-dvd
[02:30] <thaytan> catch you all later!
[02:30] thaytan ([email protected]) left irc: "See y'all"
[02:38] BBB ([email protected]) joined #gstreamer.
[02:39] BBB_ ([email protected]) joined #gstreamer.
[02:42] <LeRoutier> re BBB, i saw you DLed my test cases
[02:42] <BBB_> I didn't
[02:43] <BBB_> I will look at it earliest sunday afternoon
[02:43] <BBB_> my timing sucks right now
[02:44] <LeRoutier> someone at  condor.ucsf.edu (128.218.64.97) downloaded them tonight
[02:45] <LeRoutier> oh, k, it's ds-work, neverwind
[02:46] <ds-work> I didn't think that 'get -r *' would get so much
[02:47] <LeRoutier> there might be 300MB in riff-avi (about 200MB in riff-avi/_bad_ )
[02:47] <ds-work> depending how fat your connection is, I may have gotten all of it
[02:47] <ds-work> or most
[02:47] Action: ds-work can slurp at about 8-9 MB/s
[02:48] <LeRoutier> got an "unlimited" bw (well, limited by the 100Mb/s network card
[02:49] <ds-work> heh, I pulled 263 MB :)
[02:50] <LeRoutier> started 02:39:30, finished 02:51:29 
[02:50] <LeRoutier> in 12 minutes so 
[02:50] <LeRoutier> FTP sucks when there are several small files
[02:53] Rotty ([email protected]) left irc: Remote closed the connection
[02:54] abo_ ([email protected]) left irc: Read error: 113 (No route to host)
[02:56] <ds-work> I think I have enough stuff to fix... :)
[02:57] Action: ds-work comments about having to do this all over again with the new autoplugger
[02:57] dholmes ([email protected]) joined #gstreamer.
[02:58] <LeRoutier> ds-work, for the pad issue, yes. but for the fact that most of those videos use 4CC codes that aren't handled by gstreamer actually, it would be a one time fix
[02:58] BBB ([email protected]) left irc: Read error: 110 (Connection timed out)
[02:58] <LeRoutier> 4CC => codecs


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click