IRC Logs

*******************************************************************
[03:05] <thomasvs-skate> ds-work: did you switch sinesrc back to use double too ?
[03:06] Nick change: thomasvs-skate -> thomasvs
[03:06] Action: thomasvs just watched antitrust
[03:06] <ds-work> thomasvs: it was never changed to float
[03:07] <thomasvs> ds-work: didn't I commit stefan kost's patch to do that two days ago ?
[03:07] <ds-work> I hope not, since I commented on the bug that it was wrong
[03:08] Nick change: mathrick|afk -> mathrick
[03:09] <mathrick> thomasvs: you're skating at 3:00AM?
[03:09] <ds-work> barcelona causes people to do strange things
[03:09] <thomasvs> ds-work: hehe :) well, I remember committing, I'll check tomorrow then I guess
[03:09] <ds-work> let's just hope he wasn't naked
[03:10] <thomasvs> skating naked would be really painful
[03:10] <thomasvs> mathrick: watched a movie, too lazy to switch nicks first
[03:11] <mathrick> thomasvs: k. OTOH, 3:00AM isn't that late in Barcelona if it was summer :)
[03:12] <ds-work> ok, /me now dislikes dparams
[03:12] <thomasvs> ds-work: why ? i think they're neat
[03:13] <ds-work> they should be special pads
[03:13] <thomasvs> ds-work: I asked for control pads a long time ago, but ...
[03:13] <ds-work> as well as special object properties
[03:13] <thomasvs> do we still suck at pause/play ? if so, the clocking really needs to change
[03:13] Action: thomasvs is going to check old and new system next week
[03:14] Nick change: thomasvs -> thomasvz
[03:14] <thomasvz> later
[03:14] Action: mathrick goes to bed too
[03:14] <mathrick> nite
[03:15] Nick change: mathrick -> mathrick|sleep
[03:21] Action: sxpert wonders how gstreamer will be able to be used on handheld processors that don't have FPU support when *all* plugins use doubles
[03:21] <ds-work> not all plugins use doubles
[03:23] <sxpert> ds-work: I mean, any sort of floating point values
[03:25] <ds-work> few plugins use floating point
[03:27] <sxpert> imho, the autoplugger needs to be processor abilities aware so as to not use floating point when not absolutely necessary...
[03:27] <ds-work> elements merely need to adjust their rank
[03:28] <sxpert> ds-work: yeah, whatever ;D
[03:28] <sxpert> we need that functionality.
[03:28] <ds-work> it already works
[03:28] <sxpert> ah...
[03:30] Nick change: sxpert -> assurancetourix
[03:30] Nick change: assurancetourix -> sxpert
[03:45] iain_ ([email protected]) left irc: "leaving"
[03:49] DarkStorm ([email protected]) left irc: 
[03:52] <desrt> hah!
[03:52] <desrt> they *finally* unmasked the 1.3 series of galeon in portage
[03:54] <sxpert> lol
[04:29] ChriHJW_log ([email protected]) joined #gstreamer.
[04:30] jaiserca ([email protected]) left irc: "Terminando cliente"
[04:36] ChrisHJW_log ([email protected]) left irc: Read error: 60 (Operation timed out)
[05:11] danb ([email protected]) left irc: "moving into living room"
[05:15] NFusi0n ([email protected]) left irc: Read error: 110 (Connection timed out)
[05:32] jcsston ([email protected]) joined #gstreamer.
[05:46] thomasvz ([email protected]) left irc: Connection timed out
[05:47] thomasvz ([email protected]) joined #gstreamer.
[06:20] walters ([email protected]) left irc: "out"
[06:27] danb ([email protected]) joined #gstreamer.
[06:33] thaytan ([email protected]) joined #gstreamer.
[06:36] <thaytan> heya
[06:36] <thaytan> who's awake?
[06:40] <bitpoet> do i count?
[06:40] <thaytan> I was hoping for thomas or uraeus, but sure :)
[06:41] <bitpoet> unfortunately, i am neither of them
[06:41] <thaytan> try not to be too sad, we can't all be them :)
[06:41] <bitpoet> i guess not  :-)
[06:52] Action: thaytan is away: time for a swim
[07:04] sub_pop ([email protected]) left irc: "Client exiting"
[08:33] ChrisHJW ([email protected]) joined #gstreamer.
[08:36] walters ([email protected]) joined #gstreamer.
[09:19] Nick change: mathrick|sleep -> mathrick
[09:19] <mathrick> mornin
[09:30] mathrick_ ([email protected]) joined #gstreamer.
[09:30] mathrick ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[09:31] <thaytan> yo
[09:31] <mathrick_> yo thaytan
[09:32] Nick change: mathrick_ -> mathrick
[09:32] Nick change: mathrick -> mathrick|food
[09:35] Company ([email protected]) joined #gstreamer.
[09:35] <Company> good morning angels
[09:45] markey ([email protected]) joined #gstreamer.
[09:53] <thaytan> hi there
[09:54] <thaytan> I lost my laptop :(
[09:55] <thaytan> we got broken into last night
[09:55] <Company> shit :(
[09:55] <Company> only the laptop or lots of code, too?
[09:56] <thaytan> heh
[09:56] <thaytan> bit of both
[09:56] <thaytan> I had all the dvd stuff on there only
[09:56] <thaytan> and some speex work
[09:57] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[09:57] <thaytan> 'get contents insurance' was still on my todo list too
[09:58] <Company> :/
[09:58] <thaytan> oh well
[09:58] <bitpoet> maybe the thief will show up here asking for help
[09:58] <thaytan> our revised budget says we can buy a new one in June
[09:58] <bitpoet> if anybody wants help hacking speex and dvd we'll trace him for you  :-)
[09:59] markey ([email protected]) left irc: "bbl"
[09:59] <thaytan> or if anyone logs on as thaytan not from this address, means they've plugged it into a network somewhere ;)
[10:00] <thaytan> Company: I do remember finding last night that it is possible for current_chain to be -1 in gst_ogg_get_pad_by_pad
[10:00] <thaytan> I think it was when receiving an event
[10:01] <Company> hm
[10:01] <Company> possible
[10:01] Action: Company is good at forgetting his own code
[10:01] <thaytan> it doesn't check for it, so it died
[10:01] <thaytan> tv calls... catch you all later :)
[10:01] harshy ([email protected]) left irc: "I quit for now"
[10:04] <bitpoet> i certainly hope a bunch of people start logging on as "thaytan" now.  :-)
[10:07] <Company> yeah
[10:07] <Company> he should get a new ssh id
[10:07] <Company> i don't want funny commits to gstreamer
[10:07] harshy ([email protected]) joined #gstreamer.
[10:19] <Company> ok, now to wrap xine
[10:19] <Company> i need some light stuff to work on ;)
[10:26] <thaytan> Company: I already mailed thomas to delete the old one :)
[10:26] <thaytan> (paranoia)
[10:28] <Company> weee, i love such people
[10:28] <Company> i'd have never thought about that
[10:29] <bitpoet> wait, do we know this is the real thaytan?  maybe this is thief-thaytan, just *claiming* he had thomas delete it.
[10:30] <Company> :)
[10:30] <Company> did we even know the old thaytan is the real one
[10:30] <Company> maybe this is the real one who only stole his laptop back
[10:30] <Company> or something?
[10:30] <bitpoet> good point!
[10:31] <bitpoet> *i* am spartacus
[10:37] <Company> i need an @gnome.org or @gstreamer.net or @freedesktop.org address
[10:37] <Company> public.uni-hamburg.de is too long
[10:37] <Company> and it's not 1337 enough anyway
[10:42] <thaytan> hahahaha
[10:43] <thaytan> when I get to barcelona, I'll exchange a sekret message with thomas so we can verify in future ;)
[10:43] <Company> you'll just replace thomas
[10:48] <thomasvz> Company: I'm having erik set us up with gstreamer.net or gstreamer.org addresses
[10:48] <Company> cool
[10:50] Nick change: thomasvz -> thomasvs
[10:58] <thaytan> thomasvs: got my email?
[11:04] walters ([email protected]) left irc: "out"
[11:06] Jara[zZ] ([email protected]) left irc: Remote closed the connection
[11:07] wheels ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[11:07] markey ([email protected]) joined #gstreamer.
[11:10] Jara[zZ] ([email protected]) joined #gstreamer.
[11:18] kwm ([email protected]) joined #gstreamer.
[11:18] <Company> oooops
[11:19] <Company> wrapping xine is a bit worse than i thought
[11:21] <Company> why do audio codecs have to close the audio device, they should send EOS
[11:21] <Company> evil xine *slap*
[11:27] Shoragan ([email protected]) joined #gstreamer.
[11:31] thomasvs ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[11:37] Nick change: mathrick|food -> mathrick|away
[11:38] markey ([email protected]) left irc: "bbl"
[11:47] thomasvs ([email protected]) joined #gstreamer.
[11:52] <thomasvs> thaytan: yep.  how long do you think it will be before you have a new pc ? if possible i'd like to ask for removal and readdition at hte same time
[11:52] <thomasvs> thaytan: most thiefs don't know what an ssh key is
[11:53] Lauwenmark ([email protected]) joined #gstreamer.
[11:54] mathrick|away ([email protected]) left irc: Remote closed the connection
[12:01] iain_ ([email protected]) joined #gstreamer.
[12:01] Shoragan ([email protected]) left irc: "Leaving"
[12:03] <Company> weee
[12:03] <Company> the priorities for xine audio plugins are fun
[12:09] <iain_> hey Company 
[12:10] <Company> hey iain
[12:10] <Company> _
[12:10] <iain_> yeah, some damn bot has ain
[12:10] <iain_> and the actual owner of the nick iain keeps kicking me off
[12:10] <Company> you should register some nick
[12:11] <Company> loosers would take "therealiain", so pick something better ;)
[12:11] <iain_> I just like iain
[12:11] <Company> /nick marlin ?
[12:11] <Company> :p
[12:11] <iain_> ewww
[12:11] <Company> lene is probably taken ;)
[12:12] <iain_> probably
[12:12] <iain_> but I'm still not sure I want to go for that whole psycho obsessed stalker image
[12:14] <Company> you gotta find something else than iain or ain at least
[12:15] <iain_> but I like iain
[12:15] <iain_> I don't even like ain
[12:15] <iain_> really
[12:15] <Company> well, maybe you can get the other iain to sell his account to you :p
[12:17] <iain_> he lives in glasgow...I'll go over and kill him
[12:18] <Company> you're from northern ireland? :p
[12:22] <iain_> yes...
[12:22] <iain_> its only 49p for a flight to glasgow
[12:23] <Company> anybody bored in here?
[12:23] <Company> i need someone to convert xine formats to caps
[12:24] <iain_> I'm bored, but also not motivated :)
[12:24] Uraeus ([email protected]) joined #gstreamer.
[12:24] <Uraeus> hola
[12:25] <iain_> yo
[12:29] <Lauwenmark> Stupid question: shouldn't "gst-launch filesrc location=video.mpeg ! mpegdemux video_%02d!src { queue ! ffdec_mpegvideo ! xvideosink }" be valid to play an mpeg1 video through GStreamer ?
[12:29] <Company> not in 0.7
[12:30] <Lauwenmark> and in 0.6.4 ?
[12:31] <Uraeus> looks like a valid 0.6.x pipeline to me
[12:32] <Uraeus> but I am not sure how well developed all the ff* plugins where in 0.6, tried the mpeg2dec plugin?
[12:36] sublett ([email protected]) joined #gstreamer.
[12:44] <Company> is there a way to make ld look for duplicate symbols?
[12:45] <Company> or valgrind or whatever?
[12:45] <Uraeus> thaytan: now that we have a new 0.7 release, should we make a gst-editor release too?
[12:49] <Lauwenmark> Uraeus: Tried the mpeg2dec plugin as well. Now I'm wondering why nothing is displayed as the result of that string. Did I forget something ?
[12:50] <Uraeus> Lauwenmark: have you looked at the pipeline inside gst-launch-ext? it should contain a working mpeg pipeline you could modify
[12:51] parsley ([email protected]) joined #gstreamer.
[12:53] <Uraeus> hi parsley
[12:53] <parsley> Uraeus, howdy
[12:53] <Lauwenmark> Uraeus: Well, I got my line from the man page of gst-launch, so I supposed it was correct. but I get no visual output in the end.
[12:53] <Uraeus> Lauwenmark: do you get visual output when just trying to play the file with gst-launch-ext ?
[12:54] <Lauwenmark> Uraeus: Yes, so I'll check its content. I probably missed something, or used a wrong element in the chain.
[12:55] Shoragan ([email protected]) joined #gstreamer.
[12:56] <parsley> I'm trying to do something that seems like it should be simple: play an .ogg file from a pipe
[12:56] <parsley> I tried this: cat my.ogg | gst-launch fdsrc ! vorbisfile ! osssink
[12:57] <Uraeus> I have never used fdsrc, but your pipeline seems correct
[12:57] <parsley> just using gst-launch filesrc location=my.ogg ! vorbisfile ! osssink works
[12:57] <Uraeus> parsley: what kind of errors do you get?
[12:57] pb_ ([email protected]) joined #gstreamer.
[12:58] <parsley> hrm... this is v 0.6.0, I'm compiling 0.6.5/0.6.4plugins now
[12:58] <parsley> Uraeus, no errors, just no sound ;-)
[12:58] <parsley> it just says 'RUNNING PIPELINE'
[12:58] <Uraeus> parsley: tried enablind debug ouput?
[12:58] <parsley> Uraeus, not yet ;-)
[12:59] Action: parsley checks the manpage
[13:04] LeRoutier ([email protected]) joined #gstreamer.
[13:04] <LeRoutier> hello
[13:05] <Uraeus> hi LeRoutier
[13:05] <parsley> Uraeus, hrm... seems it's trying to 'seek' on the fd :-(
[13:05] <LeRoutier> what's going on with ffmpeg gst-plugin ? got to checkout a new module ?
[13:08] <Uraeus> LeRoutier: yes
[13:08] <parsley> Uraeus, hrm, I think maybe I want 'pipefilter'...
[13:08] Action: parsley checks it out
[13:08] <Uraeus> parsley: well hopefully it was fixed later in th 0.6.x series; never used fdsrc myself
[13:09] <LeRoutier> in which order ? gstreamer + gst-ffmpeg + gst-plugins or gstreamer + gst-plugins + gst-ffmpeg ?
[13:09] <Uraeus> LeRoutier: last one
[13:09] <LeRoutier> k, thx
[13:10] <LeRoutier> making a CVS ebuild now and testing
[13:13] <Company> parsley: i think i fixed quite some bugs wrt fdsrc / seeking in 0.7
[13:14] <Company> parsley: i wouldn'T be surprised if it didn't work in 0.6
[13:14] <Company> parsley: it works with cvs HEAD though
[13:14] <Company> i'm using it atm
[13:14] <parsley> Company, ah... cool, thanks... I'll try plugins 0.6.4, failing that, I'll move on to head
[13:15] <parsley> Company, you mean, my fdsrc pipe above is working for you?
[13:15] <Company> parsley: yeah, using HEAD
[13:16] <parsley> ok, good to know I _can_ get it to work; I'll try 0.6.4 shortly and let you know
[13:18] dolphy ([email protected]) joined #gstreamer.
[13:19] wheels ([email protected]) joined #gstreamer.
[13:19] <LeRoutier> hi dolphy, wheels
[13:19] <dolphy> yop
[13:19] <wheels> howdy
[13:19] <LeRoutier> Uraeus, works well (gst-ffmpeg compiles and installs ok), thx
[13:20] <Uraeus> LeRoutier: yup, just compiled it myself, even added a spec file to it
[13:20] <LeRoutier> yep, i see it in the changelog
[13:20] <LeRoutier> (http://cia.navi.cx/stats/project/GStreamer)
[13:21] The_Company ([email protected]) joined #gstreamer.
[13:21] <LeRoutier> i reload this page from time to time to see if something interresting comes
[13:23] Company ([email protected]) left irc: Nick collision from services.
[13:23] Nick change: The_Company -> Company
[13:24] Nick change: Kaetzchen -> alley_cat
[13:25] teuf ([email protected]) joined #gstreamer.
[13:30] <Uraeus> LeRoutier: that is a cool stats page they have 
[13:31] <LeRoutier> yep, and they also provide a rss feed for commits
[13:36] ramon ([email protected]) joined #gstreamer.
[13:36] <ramon> Hi.
[13:36] <ramon> I am here for a small support request.
[13:36] <Uraeus> hi ramon
[13:37] <ramon> If an element wants to handle events (other that state change events), it should implement GstElementClass->send_event, should it?
[13:37] <ramon> Hi Uraeus.
[13:38] <Uraeus> ramon: seen the chapter on events in the PWG?
[13:40] <ramon> No, sorry, I have not built the PWG. I should.
[13:40] <Uraeus> ramon: http://www.freedesktop.org/~gstreamer/data/doc/gstreamer/head/pwg/html/index.html
[13:40] <Uraeus> ramon: there is a chapter called : Events: Seeking, Navigation and More
[13:40] <dolphy> ramon: what events are you talking about
[13:40] <dolphy> ramon: downstream or upstream events ?
[13:41] <Uraeus> dolphy: I just played Shrek Quicktime movie in gst-player :)
[13:41] <dolphy> Uraeus: goood ! :)
[13:41] <dolphy> Uraeus: i m rewriting switch now to fix the EOS issue in totem
[13:41] <dolphy> Uraeus: my bug list is decreasing slowly
[13:42] <dolphy> but surely :)
[13:43] <ramon> dolply: upstream events now.
[13:43] <Uraeus> yup I think things are starting to fall into place
[13:43] <Uraeus> ramon: http://www.freedesktop.org/~gstreamer/data/doc/gstreamer/head/pwg/html/section-events-upstream.html
[13:44] <dolphy> ramon: ok please read the documentation pointed to you
[13:44] <dolphy> ramon: i wrote that part
[13:44] <dolphy> ramon: if you have comments or question please feedback
[13:44] <ramon> Thanks, but I do not have a DSL line, so I rather try to build it for online use.
[13:45] <dolphy> ???
[13:45] <dolphy> that s an html page
[13:46] <ramon> For offline use.
[13:46] <dolphy> just save it then 
[13:46] <ramon> I want to have it in my hard drive so that I can see it without a network connection.
[13:46] <dolphy> and moreover you can have that build gstreamer core
[13:46] <ramon> Yes, but saving all the pwg is rather complicated.
[13:46] <dolphy> this html pages are generated when building core
[13:48] <ramon> I will try using wget.
[13:48] <ramon> I am unable to build. I guess I need some stylesheets.
[13:49] <dolphy> you need basic openjade stuff
[13:49] <dolphy> for docbook tools
[13:49] <ramon> dolphy: great work. the page that you wrote is simply perfect.
[13:50] Action: dolphy blushes
[14:03] <Company> damn
[14:03] Action: Company shoots ffmpeg
[14:06] Action: Company excuses himself
[14:06] Action: Company reloads
[14:06] Action: Company shoots xine
[14:06] <Company> ooops
[14:07] <Uraeus> thomasvs: we need some clever fix for the SPEC files so they recognized the gstreamer07 packages as the heirs to gstreamer
[14:09] <teuf> Company: I'm looking at this metadata transcoding issue, id3tag sends a tag event with the proper information, but lame gets an empty tag list
[14:09] <dolphy> Uraeus: i don't think he is actually here
[14:09] <teuf> but gst-launch which uses a g_signal_connect to be notified of tags can properly read them
[14:12] <Company> teuf: hmm..., that sounds like a free issue
[14:13] <Company> gst_structure_free is eeevil ;)
[14:20] markey ([email protected]) joined #gstreamer.
[14:29] <thomasvs> Uraeus: why would we need that ?
[14:29] <ramon> Hi again.
[14:29] parsley ([email protected]) left irc: Read error: 110 (Connection timed out)
[14:29] <thomasvs> LeRoutier: order doesn't matter
[14:30] <LeRoutier> thomasvs, for what ? compilation order of gst-plugins & gst-ffmpeg ?
[14:30] <thomasvs> ramon: just get the ps or pdf ?
[14:31] <thomasvs> LeRoutier: yep
[14:31] <Uraeus> thomasvs: forget it, I solved it, I will commit my fix
[14:31] <thomasvs> Uraeus: solved what ?
[14:32] <iain_> hmm, I need to get my life sorted out
[14:33] Action: iain_ has had an epiphany
[14:33] Nick change: iain_ -> ^iain^
[14:33] <Uraeus> thomasvs: the gst-plugins rpms etc., depend on GStreamer != Gstreamer07, but I added a provides line to gstreamer.spec.in now
[14:35] <ramon> thomasvs: I have just finished downloading all the pwg with wget.
[14:38] <thomasvs> Uraeus: what exactly did you add ? I want to be sure you're not breaking the stuff I worked on
[14:38] <dolphy> thomasvs: do you know if all requested pads are correctly cleaned when the element gets disposed ?
[14:40] <thomasvs> dolphy: no, but my guess would be "no"
[14:41] <dolphy> thomasvs: hmm ok i will check
[14:43] <^iain^> what does "pro rata" mean in a job description?
[14:44] <ramon> If I set the send_event function in an element instead of in a pad, is there any important difference?
[14:44] <teuf> ^iain^: what is the full sentence ?
[14:46] <thomasvs> ^iain^: charge by the hour ?
[14:46] <dolphy> ramon: for upstream event you always set an event handler for a pad
[14:47] <dolphy> ramon: the src pad moreover
[14:47] <Uraeus> thomasvs: commit it, see the changelog it is a two line addition: Provides: gstreamer and Obsoletes: gstreamer
[14:47] <Uraeus> s/commit/commited/
[14:47] <dolphy> ramon: setting an event handler for the element might work for a single src pad element but you take a risk here
[14:47] <thomasvs> Uraeus: ok, that's wrong
[14:47] <thomasvs> Uraeus: why would you obsolete gstreamer ?
[14:48] <Uraeus> thomasvs: because if we don't there is no way to upgrade rpms between 0.7.3 and 0.7.4
[14:49] <ramon> dolphy: perhaps that section in the pwg should mention that customizing event handling is done by setting the method of GstRealPad.
[14:49] <thomasvs> Uraeus: there are no rpms of 0.7.3
[14:49] <thomasvs> Uraeus: moreover, you obsolete the 0.6 rpms this way
[14:51] thomasvs ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[14:51] <^iain^> teuf: "Salary £13,547 - £15,603 pro rata"
[14:52] <ramon> I am modifying the UDPsrc element.
[14:54] <pb_> ^iain^: is it a part-time job?  "pro rata" normally means that you will be paid an appropriate fraction of the annual salary depending on the number of hours you work.
[14:55] <^iain^> pb_: yeah I think its part time
[14:55] <^iain^> pb_: thanks
[14:56] Shoragan ([email protected]) left irc: "Leaving"
[14:58] thomasvs ([email protected]) joined #gstreamer.
[14:59] <Uraeus> thomasvs: there are gstreamer 0.7.3 rpms in Fedora, Mandrake Cooker etc.
[14:59] <thomasvs> Damn you Telefonica !
[14:59] <thomasvs> as I was saying
[14:59] <thomasvs> Uraeus: moreover, you obsolete the 0.6 rpms this way
[14:59] <Uraeus> thomasvs: but I guess breaking upgrade for these people might be better than obsoleting 0.6
[14:59] <thomasvs> Uraeus: where are 0.7.3 rpms for fedora ?
[15:00] <Uraeus> thomasvs: I thought 0.6.5 where supposed to version also 0.6?
[15:00] <thomasvs> mandrake cooker is not important for our spec file
[15:00] <Uraeus> thomasvs: fedora 2 test
[15:00] <thomasvs> Uraeus: well, that's alex's choice, and I'm going to rediscuss it with him, I told him before
[15:00] <thomasvs> Uraeus: there will be gstreamer06 rpms when 0.8 is out
[15:01] <thomasvs> Uraeus: the base gstreamer packageset should always be "the latest stable one"
[15:02] <Uraeus> hmmm, actually 0.6 is gone from fedora
[15:03] <thomasvs> probably
[15:04] <Uraeus> well the alternative fix I guess is to update all gstreamer using packages to look for gstreamer07
[15:05] Action: Uraeus gets started on that
[15:06] <thomasvs> yep, that's what I'm doing
[15:06] <thomasvs> I thought you didn't want to help out on the repo anymore ?
[15:06] Action: thomasvs is confused about Uraeus's sudden change of direction
[15:06] <Uraeus> change of direction?
[15:07] <thomasvs> yeah, last time you got angry again about fedora.us, and now you're working on packaging again
[15:07] <Uraeus> thomasvs: it has nothing to do with the repo, but I always run gstreamer installed from RPMS
[15:07] <thomasvs> that's kind of pointless for testing, no ?
[15:07] <thomasvs> it's better to test stuff we distribute; ie source tarballs or our official rpms
[15:08] <Uraeus> no, it lets me find a lot of disting and build issues which otherwise would pile up until release time
[15:08] foser ([email protected]) joined #gstreamer.
[15:08] <Uraeus> thomasvs: I always install my CVS snapshots as rpms also, this is not just about releases
[15:09] <thomasvs> well, it would make more sense to get the autobuild going again then, IMO
[15:10] <Uraeus> well this is what I have been doing since day 1 :) it is nothing new :)
[15:10] <Uraeus> I think I have done a source install once during the last 3-4 years :) the rest has always been rpms :)
[15:11] <pb_> Uraeus, thomasvs: are you guys going to FOSDEM next weekend?
[15:11] <thomasvs> pb_: I am, yeah
[15:11] <pb_> rock.
[15:11] <thomasvs> pb_: julien is too
[15:11] <thomasvs> and I think johan decided to too
[15:12] <^iain^> memo to self: Next time you email for a job application form, put your address in the email
[15:13] <pb_> doh
[15:13] <Uraeus> pb_: I am not going
[15:13] <^iain^> pb_: make a good first impression, thats my motto
[15:14] <pb_> ^iain^: sounds like the theory is good but the implementation leaves something to be desired
[15:14] <^iain^> pb_: thats not bad as a motto either
[15:15] <^iain^> pb_: my first thought was "Its for a computer job, I'm sure they'll have it as a word document"
[15:18] <Uraeus> gotta go, bye
[15:18] Uraeus ([email protected]) left irc: "Client exiting"
[15:23] ramon_ ([email protected]) joined #gstreamer.
[15:26] ramon ([email protected]) left irc: Nick collision from services.
[15:26] Nick change: ramon_ -> ramon
[15:27] <ramon> Hi. 
[15:31] <ramon> Everone away?
[15:34] <Company> thomasvs: you removed ffmpeg from gst-plugins?
[15:36] <thomasvs> Company: yep, ds prodded me
[15:36] <Company> thomasvs: the ChangeLog doesn't reflect that
[15:36] <Company> so i was a bit surprised
[15:38] <thomasvs> I sent a mail, no ?
[15:38] <thomasvs> oh, I forgot the actual comment in the changelog apparently
[15:38] <thomasvs> that's stupid :)
[15:38] <Company> probably haven'T read that yet
[15:39] <thomasvs> let me recommit the ChangeLog then
[15:39] <thomasvs> I'm still not entirely happy with the situation though
[15:39] <thomasvs> I guess I should look into pulling out the colorspace conversion code
[15:39] <Company> i thought we'd leave the ext/ffmpeg dir in gst-plugins
[15:40] <Company> and just put the stuf from gst-libs/ext/ffmpeg int gst-ffmpeg
[15:40] <thomasvs> how would htat work ? we link the ffmpeg code statically in the plugin
[15:41] <Company> oh
[15:41] <Company> i thought we build an .so
[15:42] <thomasvs> no - we did look into that once, but I don't think anyone ever really worked on it
[15:42] <thomasvs> well, it would be more useful anyway to have the ffmpeg guys just do decent releases
[15:43] <Company> yeah
[15:45] <ramon> How can the UDPSrc element annotate in the buffer the IP address of the machine that sent the packet?
[15:45] <Company> thomasvs: GST_ARCH checks for mmx by including include/mmx.h
[15:45] <Company> thomasvs: is that correct?
[15:45] <ramon> May I add metadata to buffers?
[15:46] <Company> ramon: it can't
[15:46] <ramon> I thought about adding metadata support to GstBuffer.
[16:01] <ramon> ???
[16:05] Nick change: markey -> markey|afk
[16:07] <Company> we haven't found a satisfactory way to handle "metadata" in buffers
[16:07] <Company> you could however invent your own format and transfer the ip inside the buffers data
[16:08] <ramon> I agree with Ronald that it would be counterintuitive.
[16:09] <ramon> What issues raises adding metadata to buffers?
[16:09] <Company> generic elements can't handle it
[16:10] <Company> what is data, what is meatadata?
[16:11] <ramon> I guess metadata is used when a non-specific element should do nothing with it and it is OK.
[16:11] <ramon> A generic element just passes it to the next element.
[16:12] <ramon> Perhaps it is better explained with an example.
[16:12] <thomasvs> Company: probably not - I'm not sure who wrote the detection originally
[16:12] <ramon> In my case I want to include IP address in the packets. The RTP element should use this information for better security.
[16:12] <Company> thomasvs: i'll file a bug
[16:13] <ramon> But in another application, just ignoring it is fine.
[16:16] <Company> ramon: so when i write an element that transfers data to another machine, should it keep metadata?
[16:17] <Company> ramon: and when i save it to a file/stream/anything for later?
[16:17] <Company> ramon: how should the electric fence plugin handle metadata
[16:18] <sxpert> is there a flite plugin ?
[16:18] <Company> ramon: why implement metadata so generic when just rtp needs it
[16:19] <Company> all those question need to be sorted out before we can implement metadata correctly
[16:19] <ramon> I guess this is best discussed in the mailing list.
[16:20] <Company> i think we want it some day, but not now
[16:20] <ramon> I think that we are making the mistake of discussing things in abstract. We should think about more concrete cases.
[16:21] <ramon> If you ask me, I sort of guess that metadata should be copied when you want to reproduce the stream "in general" as opposed to an specific format.
[16:22] <Company> there's also the question if metadata is only optional or might be mandatory for some formats
[16:23] <ramon> I think that the most important use of metadata is extensibility. You want to add something to the format of the buffer, but you do not want to break compatibility with existing elements.
[16:26] Action: Company thinks we should get more products and not just use "gstreamer" for everything including gst-editor
[16:31] <ramon> I don't think that this has much to do with the idea of metadata.
[16:31] Shoragan ([email protected]) joined #gstreamer.
[16:35] <Company> nope, that was a general thing
[16:41] herzii ([email protected]) left irc: "quitting..."
[16:44] <dolphy> grmbl
[16:44] <dolphy> there's an infinite loop when gstplay pipelines generates EOS
[16:45] <dolphy> it's setting EOS on source then spider then identity then ffcolorspace then on queue and then start again in source..
[16:45] <dolphy> and loop forever
[16:50] herzi ([email protected]) joined #gstreamer.
[16:57] <dolphy> hmm that's annoying
[17:05] <ramon> Company: I thought finally that UDPsrc should offer the IP address embedded in the buffer conditionally based on format negotiation.
[17:05] <dolphy> ramon: put the ip in the buffer's data
[17:06] <dolphy> ramon: through your own protocol
[17:06] <dolphy> ramon: why do you need that address though
[17:06] <Company> ramon: that sounds like a clever idea
[17:06] kmaraas ([email protected]) got netsplit.
[17:08] chris ([email protected]) joined #gstreamer.
[17:08] chris ([email protected]) left irc: Client Quit
[17:09] chris ([email protected]) joined #gstreamer.
[17:12] kmaraas ([email protected]) got lost in the net-split.
[17:13] <ramon> dolply: because in the RTP protocol one must know the IP address of each packet. Among other reasons for security.
[17:14] <ramon> And to detect collisions (two sources accidentally using the same identifier).
[17:14] <ramon> But BBB does not like the idea because it is counterintuitive.
[17:16] Nick change: markey|afk -> markey
[17:16] <ramon> I thought that gst_pad_try_set_caps was deprecated.
[17:16] <ramon> Is it?
[17:17] energy ([email protected]) joined #gstreamer.
[17:18] <dolphy> yeeeppeee
[17:18] <dolphy> fixored
[17:21] <ramon> ???
[17:21] kmaraas ([email protected]) joined #gstreamer.
[17:21] <dolphy> ramon: don't pay attention :)
[17:22] <foser> thomasvs: ping
[17:22] <ramon> Coding is fun.
[17:23] <Company> go dolphy!
[17:23] <ramon> Fun is the only thing money can't buy.
[17:23] <ramon> No, it was Fun is the one thing that money can't buy.
[17:24] <dolphy> hmm avidemux was looping forever on riff lib calls
[17:24] <Company> best things in life are free :)
[17:24] <dolphy> but gstplay pipeline is still not EOSing correctly
[17:24] <dolphy> ramon: well that's not completely true :)
[17:25] <Misirlou> coding is hard :(
[17:25] Action: desrt reads about freeqt
[17:25] <Misirlou> rewarding, though
[17:29] MadCow ([email protected]) joined #gstreamer.
[17:30] kwm ([email protected]) left irc: Read error: 60 (Operation timed out)
[17:31] Nick change: MadCow -> kwm
[17:32] stpere ([email protected]) joined #gstreamer.
[17:32] <stpere> hello
[17:32] <stpere> I need someone to point me where to look for this quick question:
[17:33] <stpere> I'm making an audio media player and it needs to use 2 different audio adapters
[17:33] <stpere> is there anything special I need to do with the plugins (init time) to get this done?
[17:35] <ramon> I don't think that you need anything special. Each element dealing with each sound adapter will have a property saying which one is its device.
[17:35] <LeRoutier> ++
[17:35] LeRoutier ([email protected]) left irc: "Leaving"
[17:39] <ramon> How should I define my custom event format and event name? Is it OK to add a new value to gstevent.h and a new struct to the enum? (I would add a generic struct size, data)
[17:40] <dolphy> stpere: 2 different audio adapters ?
[17:40] <dolphy> stpere: you mean outputing to 2 audio cards ?
[17:40] <dolphy> stpere: simultaneously ?
[17:41] <stpere> yes
[17:42] <stpere> maybe I should use threads as bin
[17:42] <stpere> my apps is trying to do the following
[17:43] <stpere> it is a DJ app
[17:43] <stpere> using timelines
[17:43] <stpere> you add tracks to the timelines
[17:43] <stpere> you have 2 timelines
[17:43] <stpere> one on each card
[17:43] <stpere> on the second you hear the first timeline+the second
[17:44] <stpere> I know, it's not clear =)
[17:45] <Company> anybody has gst-template checked out here?
[17:45] <dolphy> stpere: well simply use 2 different threads with an audio sink in each
[17:45] <dolphy> stpere: specifying the device for each one
[17:45] <dolphy> stpere: and you are done
[17:46] <dolphy> Company: <video_queue> retrieved data (nil) from queue
[17:46] <dolphy> Company: (nil) means there was a NULL data in queue right ?
[17:46] <stpere> nice :P
[17:47] <Company> dolphy: yeah
[17:47] <dolphy> Company: sweet
[17:47] Action: dolphy fixes..
[17:48] <dolphy> Company: we should put a safety check
[17:48] <dolphy> Company: what should i return from gst_queue_get when the queue was empty 
[17:48] <dolphy> Company: even after the interrupt
[17:48] <dolphy> Company: NULL ?
[17:51] <Company> dolphy: after an error it's ok to return NULL
[17:52] jimmy_dean ([email protected]) left irc: Read error: 110 (Connection timed out)
[17:52] jimmy_dean_ ([email protected]) joined #gstreamer.
[17:54] <Company> xine startup is slow
[17:54] <ramon> Is it reasonable not allowing to modify a property if an element is in PLAYING state?
[17:54] <Company> it needs about a second to come up here
[17:54] <Company> ramon: yes
[17:55] <dolphy> Company: seems there's a kind of race condition with queue
[17:55] <dolphy> Company: the gqueue sometimes keeps being empty
[17:56] <Company> dolphy: funny - maybe the cond is signalled though nothing got put into the queue?
[17:56] <ramon> Unfortunately one cannot return an error in set_property.
[17:56] <Company> ramon: just don't do anything
[17:56] <dolphy> Company: yeah i m trying to look deeper in the complete debug trace
[17:57] <dolphy> Company: seems that happens when autoplugging
[17:57] <dolphy> Company: event 14 is received
[17:57] <Company> that's a tagging event
[18:01] <dolphy> LOG      GST_DATAFLOW( 5257) gstqueue.c(644):gst_queue_get: <video_queue> locking t:0x85d9240
[18:01] <dolphy> LOG      GST_DATAFLOW( 5257) gstqueue.c(647):gst_queue_get: <video_queue> locked t:0x85d9240
[18:01] <dolphy> LOG      GST_DATAFLOW( 5257) gstqueue.c(721):gst_queue_get: <video_queue> retrieved data (nil) from queue
[18:01] <dolphy> seems it s not even checking for cond
[18:02] chris ([email protected]) left irc: Connection timed out
[18:04] <dolphy> Company: any clues?
[18:05] <Company> nope
[18:05] <Company> i'd need to look at it
[18:07] <dolphy> Company: want the trace ?
[18:07] <Company> nope
[18:07] <Company> i'm not into debugging this right now :)
[18:08] <desrt> is today a good day to cvsup? :)
[18:10] <Company> dunno
[18:10] <Company> it's not a good day to load all xine and gst plugins in the same app though ;)
[18:10] <desrt> i can't imagine many days are good for that :)
[18:10] <Company> i'm going to do that though :p
[18:11] <desrt> have you been commited lately?
[18:11] <Company> no
[18:14] energy ([email protected]) left irc: Read error: 60 (Operation timed out)
[18:19] <Company> ugh
[18:19] <Company> why does glib use RTLD_GLOBAL by default?
[18:19] Action: Company wakes ds-work
[18:19] <desrt> that is rather annoying....
[18:19] <desrt> yes... please.. give me giganticly nightmareish namespace colissions
[18:20] <Company> you can't even turn it off
[18:20] <Company> sweeeet
[18:21] <desrt> i hope you're being sarcastic
[18:21] <Company> yes
[18:21] <Company> gstreamer and xine both include a good set of libs
[18:22] <Company> and both obviously don't fix exported symbols
[18:22] jcsston ([email protected]) joined #gstreamer.
[18:22] <desrt> there needs to be symbol scoping
[18:22] <desrt> static/global isn't enough
[18:23] <Company> i'm wondering if it works when i load xine first
[18:23] <desrt> is there a way to pass a list of global symbols to ld or something?
[18:23] <desrt> ie: so you can have function shared between files in your module but not outwardly visible
[18:24] <Company> i don't think that works in Linux
[18:24] <desrt> see.. that's a big problem.
[18:24] <Company> i have no clue how that works
[18:24] <Company> but if you don't use RTLD_GLOBAL, aren't those symbols private?
[18:24] <desrt> --disable-plugins-builddir ?
[18:25] <desrt> erm
[18:25] <desrt> there obviously not vissible to the main program
[18:25] <desrt> but dlopen (or whatever) keeps a list
[18:25] <desrt> and uses it for linking with new modules that you load in
[18:26] <Company> not caring if i use RTLD_GLOBAL or not?
[18:26] <desrt> no...
[18:26] <Company> then RTLD_GLOBAL is kinda useless, no?
[18:26] <desrt> if you don't use _GLOBAL then the only way to get a symbol from the library is to explicitly use dlsym()
[18:27] <desrt> no.  it's useful.
[18:27] <desrt> say you have some module foo that needs utilityfunc() and that's in bar
[18:27] <desrt> if you dlopen( "libbar.so", GLOBAL )
[18:27] <desrt> then dlopen( "libfoo.so", whatever )
[18:27] <desrt> then libfoo will automatically use utilityfunc() from bar
[18:27] <Company> yeah
[18:28] <Company> and if i don't use global, it will use its own
[18:28] <desrt> in the context of media plugins that's probably almost never useful so you just happen to get screwed
[18:28] <desrt> yes.  but i suspect it might use global first.
[18:28] <desrt> i had a problem with this once
[18:28] <Company> yeah
[18:28] <Company> but that just means i need to load xine first
[18:28] <desrt> i had a _write function in one of my modules or something
[18:28] <desrt> and i ended up calling the libc's _write
[18:28] <Company> because xine doesn't use global
[18:29] <desrt> which royally confused the crap out of me
[18:29] <desrt> how clean is the namespace use in the gstplugins?
[18:30] <Company> inside our code it's extremely clean
[18:30] <Company> the problem are the libs we link to
[18:30] <desrt> oh
[18:31] <Company> and since xine dlopens libmad-0.14 while gstreamer dlopens libmad-0.15 I have a problem
[18:31] <desrt> does dlopen() do any dependancy loading?
[18:31] <Company> yes, it does
[18:31] <Company> but i think i can solve that problem if I load xine first
[18:32] <desrt> so in theory if you did dlopen() with RTLD_LOCAL
[18:32] <desrt> then everything would work
[18:32] <Company> because xine dlopens its own mad version non-global
[18:32] <desrt> fix glib and/or go around it :)
[18:33] <Company> going around requires having some magic "load-me-first" flag somewhere ;)
[18:33] <ramon> Thanks for help .
[18:33] <ramon> see you.
[18:33] <desrt> gcc supports constructors
[18:33] ramon ([email protected]) left #gstreamer (""No war for oil"").
[18:33] <Company> doesn'T work
[18:34] <Company> because the constructor is only called after dlopening the first lib
[18:34] <desrt> no...
[18:34] <Company> i need to make sure that dlopen(mad) happens after dlopen(xine)
[18:34] <desrt> you can do it in your main program
[18:34] <Company> yeah
[18:34] <Company> but that'd require changing all the main programs
[18:34] <desrt> hmm.
[18:34] <Company> and i want a gstxine plugin...
[18:34] <desrt> oh ya.  crap :)
[18:35] <desrt> the library can't see the main function's constructor list :)
[18:36] <desrt> ... why?
[18:36] <Company> why what?
[18:36] <desrt> gstxine
[18:36] <Company> because
[18:36] <desrt> seems sort of counter-purpose
[18:36] <Company> solves the problem of needing to hack on quite some plugins
[18:37] <desrt> a quick solution to make gstreamer useable
[18:37] <Company> i just wrap xine and don't care about the crap they use
[18:37] thomasvs ([email protected]) left irc: Connection timed out
[18:37] Action: desrt watches 'spider' turn into an alias for 'gstxine' :)
[18:38] <Company> nah
[18:38] <Company> gstreamer is much better than xine and can do things xine can't
[18:38] thomasvs ([email protected]) joined #gstreamer.
[18:38] <Company> apart from that i wrap all xine plugins and not just libxine
[18:39] <Company> so you have "xineaudiodec_win32" or "xineinput_mms" and not just "gstxine"
[18:42] <Company> oh shit
[18:42] <Company> xine uses RTLD_GLOBAL, too
[18:43] <desrt> ah
[18:43] <Company> ok
[18:43] <desrt> i thought you accidentally loaded libogg into your kernel or something :)
[18:43] <Company> i'll have to hack the xine plugin list
[18:43] <Company> that'll get really evil, but who cares
[18:43] <taaz> syncmail is still messed up sending dup diffs
[18:45] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[18:45] <desrt> ahh.  metadata reading from ogg is still nice and slow :)
[18:46] <Company> yeah
[18:46] <desrt> is there something i can do to fix it?
[18:46] <Company> probably because oggdemux reads the whole file no matter what
[18:46] <Company> sure
[18:47] <Company> optimize oggdemux :)
[18:47] <desrt> right now it does like...
[18:47] <desrt> open()
[18:47] <desrt> read_all()
[18:47] <desrt> then starts streaming the components?
[18:47] <Company> no
[18:47] <Company> it does open()
[18:48] <Company> detect_substreams()
[18:48] <Company> seek(0)
[18:48] <Company> start_playback()
[18:49] <desrt> and detect_substreams scans the whole file
[18:49] <Company> yeah
[18:49] <desrt> you figure that sort of thing would be in the header
[18:49] <Company> well, you need to find beginning and end of all streams
[18:49] <Company> and ogg doesn't tell you where they end
[18:50] <Company> so i did the easy thing and just walked the whole file
[18:50] <desrt> well
[18:50] <desrt> even in the case of the most ridiculosuy stupid fileformat
[18:50] <desrt> i could look at the caps of the sink that i'm linked to
[18:50] <desrt> and realise that i only need a single audio stream, right?
[18:50] <Company> you can probably steal ideas from vorbisfile on how to make it faster
[18:50] <Company> no
[18:51] <Company> because you construct the pads depending on the streams inside the ogg
[18:51] <desrt> great.
[18:51] <desrt> that's the thing... vorbisfile knows that it only needs a single audio stream
[18:52] <Company> you could try clever tricks like seeking to the end to see if your stream ends there and if not use the normal way
[18:52] <desrt> hmm.
[18:52] <Company> that'd solve it for vorbis
[18:52] <desrt> which is 99.9% of the case for 99% of current gstreamer users
[18:52] <Company> but i'm too lazy to implement it...
[18:52] <Company> especially because it's not that easy
[18:53] <Company> with the gstreamer architecture
[18:54] <desrt> i don't even understand how it's possible to link elements in the pipeline before playing starts
[18:54] <desrt> like.. could you do....
[18:54] <Company> you don't
[18:54] <desrt> oh.
[18:54] <desrt> makes sense then :)
[18:54] <desrt> so something like...
[18:54] <Company> gst-launch handles this for you
[18:54] <desrt> src -> oggdemux -> vorbisdec -> sink
[18:54] <desrt> then PLAY
[18:54] <desrt> that would fail?
[18:55] <desrt> because oggdemux would have no output pads
[18:55] <Company> if you do it in C, yes it would fail
[18:55] <desrt> and spider takes care of it for me
[18:55] <Company> yeah
[18:55] <desrt> making sense
[18:55] <desrt> you're not about to redesign gstreamer again are you?
[18:55] <Company> why would i do that?
[18:56] <desrt> 12:52 <Company> especially because it's not that easy
[18:56] <desrt> 12:53 <Company> with the gstreamer architecture
[18:56] <desrt> you have a tendancy to change things that you don't like :)
[18:56] <Company> :p
[18:56] <Company> no, i like that part
[18:56] <Company> it's just not that easy
[18:57] <Company> weee
[18:57] <Company> it's possible to trick xine
[18:58] <Company> though it's very hacky
[18:59] <desrt> gstxine is just a temporary shim though, right?
[18:59] <desrt> *cough*
[19:00] stpere ([email protected]) left irc: "Client exiting"
[19:00] <Company> dunno
[19:00] <Company> if xine is nice enough to use
[19:00] <Company> i could stop caring about ffmpeg :)
[19:01] <desrt> hmm
[19:02] <desrt> would it seriously screw stuff up if pads appeared and disappeared?
[19:02] <desrt> is that even possible?
[19:02] <Company> that's happening all the time
[19:02] <desrt> so just play through
[19:02] <desrt> we find a stream.. create a pad
[19:02] <desrt> stream ends... kill the pad
[19:02] <Company> yeah
[19:02] <Company> that has issues
[19:03] <desrt> like screwing up pipelines?
[19:03] <Company> no
[19:03] <desrt> i can see how it might
[19:03] <Company> like not knowing how long the pipeline is
[19:03] <desrt> you mean time-wise
[19:03] <Company> and where to seek in case of a seek
[19:03] <Company> s/pipeline/stream/
[19:03] <desrt> nod.
[19:04] <desrt> ok
[19:04] <desrt> you're right
[19:04] <desrt> i'm gonna look at vorbisfile :)
[19:04] <desrt> because ogg sucks... yet vorbisfile somehow manages to know how long a file is
[19:05] thomasvs ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[19:06] <Company> vorbisfile does some clever stuff i haven't had the motivation to copy yet ;)
[19:07] jcsston ([email protected]) joined #gstreamer.
[19:09] <desrt> wow
[19:09] <desrt> the multiple sequential streams thing is a whole other issue too
[19:09] smoke ([email protected]) joined #gstreamer.
[19:09] <desrt> it's an extremely tame case and even it would totally confuse gstreamer
[19:14] <shawarma> What happened to httpsrc? I can't find it in CVS anywhere..
[19:14] <desrt> use gnomevfssrc :)
[19:15] <shawarma> desrt: So httpsrc HAS been removed in favour of gnomevfssrc?
[19:15] <desrt> i'm not saying that.
[19:15] <desrt> i don't know.  but i use gnomevfssrc.
[19:16] <desrt>   end=_get_prev_page(vf,&og);
[19:16] <desrt>   if(ogg_page_serialno(&og)!=serialno){
[19:16] <desrt> ah.  the magic code.
[19:17] ramon ([email protected]) joined #gstreamer.
[19:18] <ramon> Hi.
[19:19] alley_cat ([email protected]) left irc: "May the Source be with you!"
[19:24] thomasvs ([email protected]) joined #gstreamer.
[19:25] <ramon> Can anyone help with some doubts about IPv6?
[19:38] <pb_> depends on what your doubts are.
[19:38] <dolphy> thomasvs: syncmail is broken
[19:38] <dolphy> thomasvs: could you fix that please ?
[19:42] <ramon> My problem is that I thought that I could use IPv6 sockets only, and for connecting an IPv4 host, use ::ffff:<ipv4> as the address.
[19:42] <dolphy> s/could/can event
[19:42] <dolphy> even
[19:42] <ramon> And so IPv4 would be just a particular case.
[19:47] <pb_> right
[19:52] <ramon> Since IPv4 deployment is scarce I am considering dropping support for it. It is not worth having IPv6 if that will break IPv4 support.
[19:53] <Company> hm, i can't trick xine
[19:54] <Company> well, i can, but that'd be really evil
[19:54] <ramon> Are you considering abandoning GStreamer for Xine :-) ?
[19:55] <Company> no, i'm thinking about wrapping xine plugins in gstreamer
[19:55] <ramon> What interesting plugins?
[19:55] <Company> why does xine insist on using compiler flags like -fbreak-everything-if-its-faster ?
[19:56] <Company> the win32 plugins for example
[19:56] <desrt> compiler flags shouldn't break things.
[19:56] <desrt> if they do it's a compiler bug
[19:56] <desrt> (unless it's something like -ffast-math)
[19:56] <ramon> GStreamer should provide its own support for that.
[19:57] <shawarma> ramon: for breaking stuff? :-)
[19:57] <Company> desrt: they break execution order which breaks debugging for mere mortals
[19:58] <desrt> oh.  yes.  debugging.
[19:58] <Company> why should i do stuff that the xine guys have already done
[19:58] <Company> ?
[19:59] <ramon> shawarma??
[20:00] <desrt> mmm.. shwarma.
[20:00] <ramon> Since Directshow is similar in concept to Gstreamer, there should be little difficulty using win32 elements in GStreamer.
[20:00] <Company> apart from the fact that directshow doesn't work in Linux
[20:00] <ramon> And there properties, ... could be, for instance, be manipulated inside gst-editor, ...
[20:00] <Company> and from the fact that directshow is more similar to xine than gstreamer
[20:02] <ramon> And I don't know if debugging the GStreamer calling xine calling avifile is going to be easy.
[20:03] <ramon> Is your concern not having to deal with loading a DLL in memory god knows how, ...?
[20:03] <desrt> sort of funny
[20:03] <desrt> even the author of vorbisfile is bitching about ogg a bit
[20:04] <Company> yeah
[20:04] <Company> ogg is a wonderful streaming format
[20:04] <Company> but that's it
[20:04] <desrt> not good for files
[20:04] <desrt> ya.  i sort of got that impression.
[20:04] <Company> though you could fix that
[20:04] <desrt> you have to pick one or the other
[20:05] <Company> like adding an info stream
[20:05] <desrt> if you know how long a stream is at the beginning of it then you can't add streams later
[20:05] <Company> you'd only add the info stream for "released" files of course
[20:06] <desrt> that's actually an interesting idea
[20:06] <desrt> because if it was there you could use it
[20:06] <desrt> and if it wasn't, then do the old thing
[20:06] <desrt> and if you don't understand it, ignore it
[20:07] <shawarma> ramon: desrt was talking about a compiler breaking stuff, and you said what gstreamer should provide it's own support for that. I found that funny, but I just woke up 10 minutes ago, so.. :-)
[20:08] <Company> man, there's quite some overlap between xine and gst
[20:08] <Company> (if you only look at code that they both include)
[20:08] <Company> i've found ffmpeg, goom and mad already
[20:09] Action: desrt waits for company to find xine's gst-compatibility plugin
[20:10] <Company> that's a bit harder to do, but should be possible, too
[20:10] <Company> yay, write a gst plugin that loads a gst plugin that loads a xine plugin that...
[20:10] <Company> (we already have that with gdk-pixbuf, so it's nothing new)
[20:18] Action: desrt 's arm falls off
[20:18] <desrt> the weekend's lead guitarist plays guitar like joey ramone
[20:18] <desrt> it's insane
[20:19] sublett ([email protected]) left irc: "I like food, food is good!"
[20:21] <pb_> ramon: why is ipv6 support breaking ipv4?
[20:25] <ramon> I would like to simplify by using IPv6 sockets and addresses only, and let the kernel do the work for me. That would be implemented by, if I see an IPv4 address, I convert it to the equivalent IPv6 address, and the kernel does the work for me.
[20:25] thomasvs ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[20:28] <sjoerd_> ramon: why would you special case one address family?
[20:28] <Company> desrt: do you know of a tool that finds symbol clashes?
[20:28] Zeenix ([email protected]) joined #gstreamer.
[20:29] <ramon> Do you pretend to code and UDPSrc client so that it uses a "general" address?
[20:30] <ramon> That would be too abstract.
[20:30] <ramon> And not quite useful.
[20:30] <Zeenix> hello
[20:31] <ramon> Hi Zeenix
[20:31] <Company> grah
[20:31] <sjoerd_> ramon: i don't know what your coding :).. that's why i asked why your special casing stuff..
[20:31] <Company> visualization plugins were the only reason why gstxine crashed
[20:43] thomasvs ([email protected]) joined #gstreamer.
[20:45] <desrt> Company; nm and perl?
[20:45] <desrt> actually
[20:45] <desrt> nm, grep, sort and uniq would do it
[20:46] <desrt> nm file1.so file2.so | grep ' T ' | sort | uniq -r
[20:46] <desrt> er.  that won't work.
[20:48] <desrt> nm file1.so file2.so | grep ' T ' | cut -f 2- -d ' ' | sort | uniq -d
[20:48] <desrt> that will :)
[20:50] <Company> desrt: thx, that's at least a bit
[20:50] <Company> desrt: i was looking for something like find-clash-tool my_app_that_crashes
[20:51] <Company> that tool would intercept dlopen calls and check that there are no clashes
[20:51] <desrt> that seems impossible :)
[20:51] <desrt> maybe there's a debugging flag to dlopen you can set
[20:51] <desrt> like an environment variable
[20:51] Action: desrt looks into it
[20:52] Lauwenmark ([email protected]) left irc: Remote closed the connection
[20:53] thomasvs ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[20:57] Action: Company shoots xine
[20:57] <Company> xine doesn't have a way to force loading a plugin
[20:58] <Company> ok, let's jump through some hoops to get iut working
[20:58] <ds> moo
[20:59] <Company> ha
[20:59] <desrt> oh great
[20:59] <Company> ds: gimme a tool to find clashing symbols
[20:59] <desrt> company; the way dlopen() takes care of symbols doesn't allow for finding clashes easily at all
[20:59] <Company> ds: or better yet make it so xine and gst don't have any of those
[21:00] <desrt> it maintains a list of symbol lookup maps
[21:00] <desrt> one for each open objec
[21:00] <desrt> it never tries to merge the tables so it never looks for dupes
[21:00] <ds> Company: nm -D + perl
[21:00] <Company> ds: ... in a running app that opens plugins
[21:01] <ds> you can't
[21:01] <desrt> ds; we already went there.  and did it with grep cut sort and uniq :)
[21:01] <ds> dlopen() doesn't have anything to do that
[21:01] <desrt> Company; i can write something for you if you feel like recompiling your libc
[21:01] <Company> you could add code to libc probably to force checking all symbols on dlopen
[21:01] <Company> uh, i have no problem recompiling my libc
[21:02] <desrt> ok then :)
[21:02] <Company> as long as it doesn't break my libc, i don't have a knoppix cd here ;)
[21:02] <Company> i'll check that your code does if (getenv (...)) ;)
[21:02] <desrt> heh
[21:03] Action: ds notes that every plugin has plugin_init()
[21:03] <desrt> if( !strcmp( getenv( "HAVE_KNOPPIX_CD" ), "1" ) { ...
[21:07] alley_cat ([email protected]) joined #gstreamer.
[21:08] <ramon> Thanks.
[21:08] ramon ([email protected]) left irc: "Terminando cliente"
[21:10] thomasvs ([email protected]) joined #gstreamer.
[21:18] alley_cat ([email protected]) left irc: "May the Source be with you!"
[21:24] Action: desrt builds glibc for first test run
[21:35] alley_cat ([email protected]) joined #gstreamer.
[21:43] thomasvs ([email protected]) left irc: Read error: 60 (Operation timed out)
[21:43] <desrt> ...still building...
[21:45] sxpert_ ([email protected]) joined #gstreamer.
[21:47] md` ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[21:49] estesoyyo ([email protected]) joined #gstreamer.
[21:59] sxpert ([email protected]) left irc: Read error: 110 (Connection timed out)
[22:00] Nick change: sxpert_ -> sxpert
[22:00] estesoyyo ([email protected]) left #gstreamer ("Usa Jabber, es libre").
[22:00] thomasvs ([email protected]) joined #gstreamer.
[22:06] md` ([email protected]) joined #gstreamer.
[22:09] Nick change: markey -> markov
[22:09] <desrt> ok... so great
[22:09] <desrt> i still have NO idea how glibc works
[22:11] Nick change: markov -> markey
[22:18] stpere ([email protected]) joined #gstreamer.
[22:19] <stpere> hi, found a broken link on your website
[22:20] <stpere> Developers section
[22:20] <stpere> link to documentation page
[22:21] <thomasvs> post it
[22:25] <foser> thomasvs: docs build still broken here..
[22:26] <foser> thomasvs: and whats the exact rationale behind the current version/non-versioned binaries ?
[22:27] <taaz> versioned ones are parallel installable, non versioned ones are not ;)
[22:27] Action: taaz guesses thomasvs has other opinions on that...
[22:28] <thomasvs> foser: still no clear bug report in bugzilla about it :)
[22:28] <thomasvs> taaz: no
[22:28] <thomasvs> taaz: that's actually exactly how I think about it
[22:28] <foser> thomasvs: true... sort of because i hacked around it to work.. part of it is DESTDIR stuff which is not in the docs root *.mak
[22:29] <foser> thomasvs: other part is that it wants to install HTML when there is no html (HTML_DAT not set) and then it does 'cp -r $(HTML_DAT) <dest dir>'
[22:31] stpere ([email protected]) left irc: "Client exiting"
[22:32] ChrisHJW ([email protected]) left irc: Read error: 110 (Connection timed out)
[22:33] <desrt> no wonder glibc sucks so badly
[22:33] <desrt> if i was a glibc-dev i'd neve want to work on it
[22:34] <desrt> it takes so long to actually accomplish anything
[22:45] <desrt> and i will NEVER understand why it's fashionable to put a 200-line function in a .h file
[22:50] <taaz> everything should clearly be put in a .py file ;)
[22:50] <sxpert> desrt: because you are not a glibc dev
[22:51] <thomasvs> foser: so do a full bug report, or I can't do anything about it
[22:52] <thomasvs> the DESTDIR stuff works since we build rpms
[22:52] <foser> thomasvs: what pack is needed to build html docs ?
[22:54] <desrt> screw this.  i give up.
[22:54] <desrt> it's the impossible task
[22:57] <desrt> after that ordeal i want to investigate if i can successfully run a uclibc system :)
[22:58] <sxpert> desrt: what were you trying to achieve ?
[22:59] <desrt> debugging output for symbol clashes caused by dlopen and RTLD_GLOBAL
[23:01] <sxpert> ah... uh... that's over my head
[23:02] <desrt> ya
[23:02] <desrt> me too, it seems
[23:09] jcsston ([email protected]) left irc: Read error: 110 (Connection timed out)
[23:20] somex1 ([email protected]) joined #gstreamer.
[23:41] ramon ([email protected]) joined #gstreamer.
[23:41] <ramon> Hmm BBB is not here.
[23:41] <Company> nm shows how evil gstreamer still is
[23:42] <ramon> Why?
[23:42] <Company> all the p?lugins i have installed export 675 functions that are not in the gst namespace
[23:43] <ramon> By the way, I am removing gst_udpsrc prefix from static functions. I guess it is OK.
[23:51] <sxpert> ramon: bbb has no internet access, and laptop ist kaputt, read planet.gnome.org ...
[23:58] jcsston ([email protected]) joined #gstreamer.
[23:59] Miko5881 ([email protected]) joined #gstreamer.
[00:00] --- Sun Feb 15 2004
[00:10] <Company> ds: ping
[00:10] Uraeus ([email protected]) joined #gstreamer.
[00:11] <Uraeus> ello ello
[00:11] <ramon> Hi Uraeus
[00:11] <Uraeus> ramon: so how goes the ongoing RTP effort?
[00:12] <Zeenix> hello Uraeus
[00:12] <ramon> The RTP part is mostly stopped. I have code that is fairly complete but untested. 
[00:12] <ramon> Missing is report generation.
[00:12] <Uraeus> ramon: so why is it stopped if it is nearly complete?
[00:12] <Uraeus> hey Zeenix
[00:12] <ramon> Althought the data structures are there.
[00:13] <ramon> Before finishing it, I am hacking UDPSrc, so that I can test what I have.
[00:13] <ramon> In addition, my Ph D Thesis is taking a lot of time and effort.
[00:13] <Uraeus> ramon: phd thesises seem to tend to do that :)
[00:16] foser ([email protected]) left irc: "[ I want to believe ]"
[00:17] <taaz> phd proposals do too
[00:18] <taaz> Company: 675? ouch.
[00:18] <Company> taaz: there's some statically linked libs in there though
[00:18] <Company> taaz: like modplug for example
[00:21] <Uraeus> 675 what?
[00:22] kwm ([email protected]) left irc: "Leaving"
[00:22] <Company> exported functions that are not in the gst namespace
[00:22] <Company> and are therefore potential symbol clashes with other libs
[00:23] <ramon> A quick workaround is to remove all unnecessary exported symbols from .so files. Which is probably correct anyway.
[00:24] <ramon> Apart from plugin_init, nothing should be exported.
[00:24] <Company> go ahead
[00:24] <Company> we'll take patches
[00:24] <ramon> Ugh, the automake is difficult for me. But let me try.
[00:24] <Company> (if it were that easy, we'd already done it)
[00:25] <taaz> it is about that easy
[00:25] <Company> then i'm missing something
[00:27] Uraeus ([email protected]) left irc: "Client exiting"
[00:27] <ds> Company: pong
[00:28] <taaz> isn't there a ld flag to specify the symbols or a file with the symbols to export?
[00:28] <ramon> Yes.
[00:29] <Company> afaik that doesn't work on Linux
[00:29] <Company> ds: if i have a   while (TRUE) { gst_caps_append (caps, get_caps_from_id (i++); }  loop
[00:30] <Company> ds: how do i initialize caps correctly? gst_caps_new_empty() ?
[00:30] <ds> yeah
[00:31] markey ([email protected]) left irc: "nite"
[00:32] <Company> k
[00:32] <Company> goom is the offender with xine
[00:32] <Company> if i remove xines tvtime and goom plugins, it works
[00:32] <taaz> What is "siddisable" for?  gst/gstcpuid_i386.s:siddisable:
[00:33] <taaz> is that some magic?  it looks like nothing uses it
[00:33] <ds> 'doesn't work in user mode' would be a clue
[00:34] <taaz> a clue that i have no clue
[00:37] <ramon> Let me test it.
[00:38] <thomasvs> foser: stuff like docbook, xsltproc, a decent catalog setup
[00:38] <thomasvs> oh, gone
[00:42] sub_pop ([email protected]) joined #gstreamer.
[00:44] walters ([email protected]) joined #gstreamer.
[00:45] Miko5881 ([email protected]) left irc: Remote closed the connection
[00:45] alley_cat ([email protected]) left irc: "May the Source be with you!"
[00:46] alley_cat ([email protected]) joined #gstreamer.
[00:49] <ramon> Company: libtool has -export-symbols-regex
[00:49] <ramon> Although it might not work under Linux
[00:50] ChrisHJW ([email protected]) joined #gstreamer.
[00:52] <ramon> Under Linux une must use linker scripts.
[00:53] <ramon> Company?
[00:54] Shoragan ([email protected]) left irc: "Leaving"
[00:54] <Company> here
[00:54] <Company> and i know that it doesn'T work under Linux
[00:54] <ramon> I found how to restrict  exports.
[00:54] <ramon> No.
[00:54] <ramon> It is another method: using version scripts.
[00:55] <ramon> The simple method of Windows is not supported on Linux, probably because under Linux libraries can have versions.
[00:57] <Company> if anyone finds a portable way of making plugins just export the plugin_init stuff and nothing else (apart from API they officially export) i'm all for it
[01:05] Zeenix ([email protected]) left irc: "The whole problem with the world is that fools and fanatics are always so certain of themselves, but wiser people so full of 
[01:16] taaz ([email protected]) left irc: Read error: 110 (Connection timed out)
[01:17] Kaetzchen ([email protected]) joined #gstreamer.
[01:18] ramon_ ([email protected]) joined #gstreamer.
[01:19] ramon ([email protected]) left irc: Nick collision from services.
[01:19] Nick change: ramon_ -> ramon
[01:19] wheels ([email protected]) left irc: Read error: 110 (Connection timed out)
[01:20] Company ([email protected]) left irc: Remote closed the connection
[01:22] wheels ([email protected]) joined #gstreamer.
[01:22] alley_cat ([email protected]) left irc: Read error: 110 (Connection timed out)
[01:42] taaz ([email protected]) joined #gstreamer.
[01:42] taaz ([email protected]) left irc: Remote closed the connection
[01:43] taaz ([email protected]) joined #gstreamer.
[01:44] smoke ([email protected]) got netsplit.
[01:44] ramon ([email protected]) left #gstreamer (""No war for oil"").
[01:50] smoke ([email protected]) got lost in the net-split.
[02:09] ChrisHJW ([email protected]) left irc: Read error: 110 (Connection timed out)
[02:23] tvon ([email protected]) joined #gstreamer.
[02:23] <tvon> Bindings links on the new website are 404 enabled: http://www.freedesktop.org/~gstreamer/bindings
[02:24] <tvon> Are there an gst-sharp releases/packages or should I go for cvs?
[02:26] <tvon> eek...I see no gst-sharp in cvs...though I've heard tell of such code...
[02:27] <tvon> 404: http://www.freedesktop.org/~gstreamer/docs/
[02:27] <tvon> I think its just a bad link on http://www.freedesktop.org/~gstreamer/dev/
[02:28] <tvon> it links to there as opposed to s/docs/documentation/
[02:35] <tvon> Also..would using gstreamer soley to get information from mp3 and ogg tags/comments be severe overkill?
[02:35] <tvon> I need to run, but I read the scrollback...ttyl
[02:36] <walters> grrr.  esd sucks
[02:45] ^iain^ ([email protected]) left irc: "leaving"
[02:48] pb_ ([email protected]) left irc: "Client exiting"
[02:50] hallibaby ([email protected]) joined #gstreamer.
[02:50] hallibaby ([email protected]) left irc: Client Quit


-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click