IRC Logs

*******************************************************************
[03:01] <Zeenix> Company: i always wondered if these games also have fans
[03:02] gphat ([email protected]) joined #gstreamer.
[03:03] <gphat> what's the proper way to discern the length (time) of something you are playing w/gstreamer? 
[03:03] <gphat> like, say, an mp3 ;)
[03:08] Spundun ([email protected]) left irc: Read error: 110 (Connection timed out)
[03:09] <ds-work> gphat: gst_element_query(), typically on the sink element
[03:10] <gphat> ds-work: thanks!
[03:11] <gphat> ds-work: ah, lots of nice stuff in there
[03:25] 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 
[03:41] <gphat> The example in the manual (and in the source) called 'queue' doesn't work.  Spits the following:
[03:41] <gphat> (process:5460): GStreamer-WARNING **: bin "bin" is not the managing bin, can't be iterated on!
[03:44] <gphat> I don't really know what the point of the example is, but I was trying to make my little app fill a queue and allow a thread to read the queue, and I get the same error ;)
[03:47] <ds-work> gphat: could you file a bug please?
[03:48] <gphat> sure
[03:48] Action: gphat skitters off go b.g.o
[03:50] BBB ([email protected]) joined #gstreamer.
[03:51] <gphat> ds-work: bugged
[03:52] <taaz> wheels: ping?
[03:52] <wheels> taaz: pong -- but fading fast  :-)
[03:52] <taaz> just wondering about how to make arts detect on debian boxes
[03:53] <taaz> i see that patch you had a while ago for kio stuff but i'm having issues with arts as well
[03:53] <wheels> taaz: Yeah, well, it's more than a 2 minute explanation, but at some point I'll bug you to see where it puts the stuff and see if I can fix up the configure check.
[03:54] <wheels> Hmm, actually arts has an artsc-config that could be used...
[03:54] <wheels> taaz: But yeah -- bug me tomorrow or so and I'll see what I can do.  I've got a meeting in 6 hours and sleep somewhere inbetween.  :-)
[03:54] <taaz> wheels: artsc-config is the problem ;)
[03:55] <taaz> it returns /usr/include/kde/artsc but files are in /usr/include/kde/arts
[03:55] <taaz> which is probably right based on the name but it doesn't do what needs to be done
[03:55] Action: wheels has been in a bug fixing contest of sorts with one of the other KDE guys for the last couple of days that's worn him out.  We're both up to 35 bugs in the last 4 days.  :-)
[03:55] <BBB> damn, and I get tired from fixing 5 or so per day
[03:55] <taaz> ok, can wait till tomorrow
[03:56] <wheels> taaz: Hmm, yeah, same here actually...
[03:56] <wheels> BBB: 35 bugs, 52 commits since Friday.  :-)
[03:56] <BBB> get a life :p
[03:56] <taaz> wheels: i just don't know where to add the fix without the obvious -Iblah appending hack
[03:57] <taaz> err, i don't know what the fix is either ;)
[03:57] Spundun ([email protected]) joined #gstreamer.
[03:57] <wheels> taaz: Yeah, I'm sure there's a hack to be found.  :-)
[03:58] <taaz> oh well, debian arts package shall remain empty till then
[04:00] <wheels> taaz: yeah, really -- remind tomorrow and I'll get to it...
[04:02] <taaz> ok, one (i think) issue left for the plugin debs.  what to do with colorspace stuff
[04:02] <taaz> i put all 3 plugin libs in -colorspace but i don't like that
[04:03] <taaz> colospace and ffmpegcolorspace are lib independent so could move to -misc easily enough
[04:03] <taaz> which would seem to mean i could just call the -colorspace package -hermes
[04:04] <taaz> and then add complexity by providing gstreamerX.Y-colorspace virtual package by both -misc and -hermes (and others in the future if there ever are any)
[04:04] <taaz> i'm making this too hard ;)
[04:05] <taaz> anyone have any opinions on this?
[04:24] smoke ([email protected]) got netsplit.
[04:30] smoke ([email protected]) got lost in the net-split.
[04:35] walters ([email protected]) joined #gstreamer.
[04:35] gphat ([email protected]) left irc: "Leaving"
[04:58] Company ([email protected]) left irc: Read error: 110 (Connection timed out)
[05:09] Nick change: je4d_ -> je4d
[05:32] <Spundun> hi guys.... whats the tool that select i/o plugins.. called?
[05:33] <BBB> gnomevfs?
[05:33] <Spundun> BBB: what?
[05:34] <BBB> gnomevfs is a gnome tool for file I/O
[05:34] <BBB> or don't you mean that?
[05:34] <BBB> do you mean a gst- tool?
[05:34] <BBB> gst-inspect? gst-launch?
[05:34] <BBB> gst-editor?
[05:35] <Spundun> sorry :) i meant whats the tool to select source and sink? (sorry just configured xmms yesterday so used that term :D)
[05:35] <Spundun> so which on e is it?
[05:37] <BBB> 'select'?
[05:37] <BBB> what do you want to do?
[05:37] <BBB> gstreamer-properties?
[05:37] <Spundun> yeah select...
[05:37] <BBB> (that selects outputs/inputs in totem and rhythmbox)
[05:37] <Spundun> okkey so gstreamer-properties is the ans.. then thats good cause thats consistent with what #rhythmbox guys said
[05:38] <Spundun> so then the problem is that I dont have gstreamer-properties binary installed on my machine
[05:38] <Spundun> while I have gstreamer and gst-plugins both installed
[05:38] <Spundun> may be I just cant find it...
[05:38] Action: Spundun on mac... Fink
[05:41] <walters> Spundun: it's in gnome-media
[05:43] <Spundun> ouch!.. god knows how many more packages to install ... its a long long way from my first mp3 in RB :)
[05:46] <Spundun> thanx walters... i dont think I could have figured that one out on my own in this life
[05:46] <BBB> don't you have GNOME? :)
[05:47] <BBB> bedtime...
[05:47] <Spundun> I am on osx/panther... just installing gnome 2.6 greedily..
[05:47] BBB ([email protected]) left irc: "zzzzzz"
[05:47] <Spundun> oops... 
[05:58] spyder482 ([email protected]) left irc: "Leaving"
[06:32] walters ([email protected]) left irc: Client Quit
[06:42] gimo_ ([email protected]) joined #gstreamer.
[06:54] ChrisHJW ([email protected]) joined #gstreamer.
[06:57] <Spundun> gstreamer 0.8 /rhythmbox 0.7 works on mac osx... 
[06:57] <Spundun> but..... not smoothly
[06:58] <Spundun> everytime I do anything with my gui... I get a glitch
[06:59] <Spundun> like switching windows and all
[07:00] <Spundun> anybody got any clue about this? 
[07:00] <Spundun> ofcourse iTunes works without a problem so its not a hardware problem
[07:07] <benow> no midi controllable parms for elements, yet, are there?
[07:07] <benow> (not that i've had much luck with my controller under alsa...)
[07:26] gimo_ ([email protected]) left irc: Read error: 110 (Connection timed out)
[07:34] Action: Spundun still looking for help/clue regarding the gstreamer glitches...
[07:35] <Spundun> the glitches happen sometimes with a simple mouse move
[07:37] <benow> could be it needs a buffer somewhere in the pipeline, but I'm guessing it's something more.... perhaps try back in a few (6-8?) hrs, where more ppl will be alive.
[07:38] <benow> erm, active, even ;)
[07:40] <Spundun> benow: :).. I think it might be just a buffer thing.... cause on top rhythmbox is not showing any spectacular resource hogging
[07:41] <Spundun> okkey i will try tomorrow morning
[07:48] Surphaze ([email protected]) joined #gstreamer.
[07:57] Spundun ([email protected]) left irc: "Leaving"
[08:15] thomasvs ([email protected]) left irc: Read error: 60 (Operation timed out)
[08:20] ChrisHJW ([email protected]) left irc: Read error: 54 (Connection reset by peer)
[08:28] thomasvs ([email protected]) joined #gstreamer.
[08:59] Surphaze ([email protected]) left irc: "Client exiting"
[09:01] wheels ([email protected]) left irc: Remote closed the connection
[09:01] kmaraas ([email protected]) left irc: Client Quit
[09:10] sub_pop ([email protected]) left irc: "Client exiting"
[09:13] sub_pop ([email protected]) joined #gstreamer.
[09:26] markey ([email protected]) joined #gstreamer.
[09:29] Shoragan ([email protected]) joined #gstreamer.
[09:46] bitshifter ([email protected]) joined #gstreamer.
[09:51] bilboed_ ([email protected]) joined #gstreamer.
[10:04] bilboed ([email protected]) left irc: Read error: 110 (Connection timed out)
[10:12] dolphy ([email protected]) joined #gstreamer.
[10:13] ChrisHJW_log ([email protected]) left irc: Read error: 110 (Connection timed out)
[10:15] ploum ([email protected]) joined #gstreamer.
[10:31] jdahlin ([email protected]) joined #gstreamer.
[10:44] kmaraas ([email protected]) joined #gstreamer.
[10:44] tlg ([email protected]) joined #gstreamer.
[10:45] Rotty ([email protected]) joined #gstreamer.
[10:48] thomasvs ([email protected]) left irc: Read error: 110 (Connection timed out)
[11:05] thomasvs ([email protected]) joined #gstreamer.
[11:09] Nick change: markey -> markey|afk
[11:35] Company ([email protected]) joined #gstreamer.
[11:42] md` ([email protected]) left irc: Read error: 60 (Operation timed out)
[11:44] <Company> good morning angels
[11:48] <thomasvs> morning charlie
[11:48] Action: jdahlin shrugs
[11:49] pb_ (~pb@2002:3e03:42cc:1:20b:6aff:fe19:f290) joined #gstreamer.
[11:49] <tlg> Hi company, I might have a bug for you ...
[11:53] <lilo_booter> i'm starting to investigate integrating the mlt framework with gstreamer - anyone care to give me a few pointers on how to approach this? there are two components which i would like to develop - one being a means for gstreamer to use load our xml docs and another being the ability to load a serialised form of a gstreamer graph in mlt... i don't know much about gstreamer, and would be grateful for a couple of pointers
[11:54] <Company> tlg: go ahead
[11:54] <Company> ... and tell me
[11:55] <tlg> (;
[11:55] <tlg> so I was trying yesterday to fetch id3 infos
[11:56] <tlg> the thing is that when I put the id3tag after a spider it doens't show the tags anymore
[11:56] <thomasvs> lilo_booter: for one, graphs in gstreamer can already be serialized to xml
[11:56] <thomasvs> lilo_booter: and, the other way around
[11:57] <thomasvs> lilo_booter: but I'm assuming you want to render edit scripts in gstreamer, for hte first part, right ?
[11:57] <lilo_booter> first part would be that, yes
[11:58] <lilo_booter> probably easiest to focus on getting gstreamer to load and play our format first i think
[11:58] <tlg> this works : gst-launch -t filesrc location=01-a_perfect_circle-the_package-srpmp3.mp3 ! id3tag ! fakesink
[11:58] <tlg> this doesn't: gst-launch -t filesrc location=01-a_perfect_circle-the_package-srpmp3.mp3 ! spider ! id3tag ! fakesink
[11:59] ChrisHJW_log ([email protected]) joined #gstreamer.
[12:00] <Company> tlg: yeah, i need to seriously kick id3tag a bit
[12:00] <Company> tlg: and replace it by id3demux/id3mux
[12:01] <lilo_booter> thomasvs: the interface to our 'producer' api is basically, init, seek, get_frame and close - from a frame, get_image and get_audio... the get_image will return a yuv422 image of the requested width and height, get audio will return a set of samples of the requested frequency, channels and number of channels.. 
[12:01] <tlg> ok ...
[12:01] <lilo_booter> thomasvs: i feel that should be fairly straighforward to integrate, right?
[12:03] <Company> tlg: problem is that i'm lazy so i didn't get around changing that yet
[12:03] <Company> tlg: it's a boring thing to do... :/
[12:03] <tlg> (:
[12:04] <jdahlin> thomasvs: have you seen this http://bugzilla.gnome.org/show_bug.cgi?id=139909 ?
[12:04] sublett ([email protected]) joined #gstreamer.
[12:07] <thomasvs> lilo_booter: yes and no :) basically you want to give it an edit list and ask for a given frame right ?
[12:07] md` ([email protected]) joined #gstreamer.
[12:08] <thomasvs> jdahlin: yeah, ensonic said yesterday he was going to do a patch, then he said "never mind, I'll just install the right versions"
[12:10] <jdahlin> thomasvs: maybe we should enforce the right versions
[12:11] <lilo_booter> thomasvs: not really - the gst plugin will load our xml format as a normal file in your framework - you needn't care aout the internal make up at all
[12:11] <thomasvs> jdahlin: sure, but I'm too lazy and uncaring to work on that :)
[12:12] <thomasvs> lilo_booter: hang on - the xml format you provide is an edit list containing descriptions of media files, timelines, and effects, right ?
[12:14] <lilo_booter> thomasvs: yes - but to you that'd be a black box - i would provide a 'source_element' which parses that
[12:14] <lilo_booter> sorry - trying to match to your terminoligy here...
[12:14] <lilo_booter> it also provides a demuxer
[12:14] <thomasvs> lilo_booter: so what framework would be doing the conversion from your xml file to video/audio output ?
[12:15] <lilo_booter> yes
[12:15] <lilo_booter> sorry 
[12:15] <lilo_booter> mlt 
[12:15] <thomasvs> ah, ok :)
[12:15] <thomasvs> now I get it :)
[12:15] <lilo_booter> heh - ok :-)
[12:15] <thomasvs> so you just want mlt to provide data sinks to the rest of gst
[12:15] <lilo_booter> yes
[12:15] <thomasvs> I thought you wanted to use gst to render it :) silly me
[12:15] Action: thomasvs puts plans for world domination back in the closet
[12:15] <lilo_booter> heh - no - that would be to hit or miss :-)
[12:15] <thomasvs> lilo_booter: so yes - then you're on the right track.
[12:16] <jdahlin> thomasvs: what docbook version do you have?
[12:16] <lilo_booter> ok - so what i need i guess is the closest match to this existing in your code
[12:17] <thomasvs> jdahlin: 4.2 and 4.1.2
[12:17] <lilo_booter> basically, it would be a demuxer of some sort i guess? mebbe the ffmpeg one? or is there something simpler?
[12:17] <thomasvs> lilo_booter: something like avidemux would be best I guess
[12:17] <thomasvs> or mpegdemux
[12:17] <thomasvs> not the ffpmeg one, too complicated a setup
[12:18] <lilo_booter> ok
[12:19] <lilo_booter> is my goal clear though? basically, mlt will become the framework on which kino will run... i'd like to make kino projects directly loadable in gstreamer
[12:20] <lilo_booter> and ditto, i would like serialised gstreamer graphs loadable in kino (so they behave as files to us)
[12:21] <lilo_booter> right now, i'm not too concerned about sharing fx or allowing editing within the serialised forms of either - just simple specification of in/out points to allow arrangement on timelines is the focus
[12:21] <thomasvs> ok, so the second point is a bit unclear to me
[12:22] bitshifter ([email protected]) left irc: Read error: 60 (Operation timed out)
[12:22] <thomasvs> because, basically, gstreamer graphs are states of a pipeline
[12:22] <thomasvs> not really timelines
[12:22] <lilo_booter> hmm - but they're seekable aren't they?
[12:23] <thomasvs> well, yeah.
[12:24] <thomasvs> if you create graphs that end in a muxer, then I guess you have what you want.
[12:24] <thomasvs> so you want to make them file abstractions, really ?
[12:24] <lilo_booter> yes - pretty much
[12:24] <thomasvs> that would probably work, yeah.
[12:24] <lilo_booter> put it this way - say you have a bit of video, and gstreamer provides an effect which you want on that which kino doesn't - you could use gst-editor to plug the two together and then load that in kino
[12:25] <lilo_booter> and ditto - say mlt/kino does something which you don't, you can loop back into the gst-editor later on to add the final touches - iyswim
[12:25] <thomasvs> that means you have to link part of kino to gst, or provide a component hooking them
[12:26] <lilo_booter> that's right - the second part of this would involve developing an 'mlt producer' which loads gstreamer stuff
[12:27] <lilo_booter> so, assuming we can bend the gstreamer api to ours and vice versa, we can, more or less, seamlessly migrate stuff between the two projects
[12:28] <thomasvs> yeah, would be nice
[12:28] <thomasvs> the real question is, when are you going to rewrite kino using gst ? :)
[12:28] <lilo_booter> not exactly world domination for either of us - but would provide a nice compromise i think :-)
[12:28] <thomasvs> yeah, exactly
[12:29] <thomasvs> and I think kino is nice anyway
[12:29] <lilo_booter> heh - we'd do that only if mlt screws up ;-)
[12:29] <lilo_booter> thanks :-)
[12:30] <lilo_booter> this integration is going to be quite a departure from the current code base... hopefully we'll retain some of the look and feel and widgets, but internally.. hmm.. not sure how much we can reuse
[12:31] <lilo_booter> anyway - i really would like to loosely couple the two projects in this way... feel it would be best for everyone
[12:32] <lilo_booter> we're about to release mlt and i have a release candidate tar ball if you want to try it (but don't feel you have to)
[12:35] bitshifter ([email protected]) joined #gstreamer.
[12:35] <thomasvs> lilo_booter: does it have a lot of dependencies ?
[12:36] <lilo_booter> very few - all dependencies are optional
[12:37] <lilo_booter> http://cvs.sourceforge.net/viewcvs.py/mlt/mlt/docs/install.txt?rev=1.5&view=markup
[12:38] ensonic ([email protected]) joined #gstreamer.
[12:38] <lilo_booter> you can ignore mainconcept and bluefish - those components aren't available in cvs or the releases
[12:38] <ensonic> hi
[12:41] <thomasvs> bluefish ?
[12:41] <lilo_booter> the SDI hardware used by our customer (it's mentioned in the dependencies section of the url above)
[12:42] <thomasvs> oh, sorry.  you must get that a lot :)
[12:43] <thomasvs> what's the relation between mlt and kino ?
[12:43] <thomasvs> ensonic: your doc bug from yesterday, what do we do with it now ?
[12:43] <lilo_booter> heh
[12:44] <lilo_booter> relationship is that we are currently in the process of redesigning kino to use mlt as its multitrack, multiformat, fx and general plugin mechanism...
[12:44] <ensonic> thomasvs: I'll contacted suse so that they update their dist. (the tools they install are now about 3yrs. old) - for me it would be okay to close the report (won't fix), fixing it would mean grepping through the --version output and using perl -e to check the version ...
[12:44] desrt ([email protected]) left irc: Remote closed the connection
[12:44] desrt ([email protected]) joined #gstreamer.
[12:46] <lilo_booter> theory is that kino will move into the 'what is expected of an nle' direction rather than the 'playlist editor with fx bolted on as an afterthought' kinda thing that it is now
[12:49] <lilo_booter> our main focus is getting paid for this stuff though, and for that, the direction of mlt is mainly in the tv broadcasting domain (kino is a betwixt contract advertising vehicle i guess - obviously, the more it gets used, the better mlt becomes...)
[12:52] smoke ([email protected]) joined #gstreamer.
[13:04] <lilo_booter> anyway thomasvs - have to be getting on - thanks for your time :-) - i'll have a poke around with demuxers soon and let you know how i get on with the first part of this
[13:06] <thomasvs> lilo_booter: no worries
[13:09] Company ([email protected]) left irc: Remote closed the connection
[13:18] Nick change: markey|afk -> markey
[13:23] Nick change: harshyOut -> harshy
[13:26] ensonic ([email protected]) left irc: Read error: 60 (Operation timed out)
[13:43] Nick change: markey -> markey|afk
[14:15] kmaraas ([email protected]) left irc: Client Quit
[14:18] Nick change: markey|afk -> markey
[14:18] LeRoutier ([email protected]) joined #gstreamer.
[14:18] <LeRoutier> hello
[14:31] bitshifter ([email protected]) left irc: Read error: 60 (Operation timed out)
[15:00] kmaraas ([email protected]) joined #gstreamer.
[15:22] mxpxpod ([email protected]) left irc: "Ex-Chat"
[15:29] ensonic ([email protected]) joined #gstreamer.
[15:36] jonathang ([email protected]) joined #gstreamer.
[15:39] <jonathang> hello everybody
[15:39] <jonathang> when i try to launch gst-editor from cvs, i get "home/jon/build/gstreamer/cvs/gst-editor/src/.libs/lt-gst-editor: relocation error: /home/jon/build/gstreamer/cvs/gst-editor/libs/gst/editor/.libs/libgsteditor.so.0: undefined symbol: editor_debug_cat"
[15:40] <jonathang> any ideas?
[15:43] <LeRoutier> you run gst-editor from its local CVS tree ? won't it be better to do it once installed where it belongs ?
[15:46] <jonathang> i run it using src/gst-editor from its local CVS tree, that's right
[15:46] <jonathang> but gst-player works like that
[15:46] <jonathang> and i think gst-editor used to work like that too
[15:54] <jonathang> is there a way to install gst-editor if it's using uninstalled cvs version of gstreamer & gst-plugins?
[15:56] <LeRoutier> i really don't know, never tried. i always use current CVS installed system wide, it's not that risky
[15:56] <LeRoutier> it's quite stable (more than 0.8.0 tarballs)
[15:57] Company ([email protected]) joined #gstreamer.
[15:57] <LeRoutier> h Company
[15:57] <LeRoutier> +i
[15:58] <Company> hi
[15:58] Action: Company is sitting in the garden wondering how long his wireless connection will survive
[15:58] <jonathang> ok, i'll do that
[16:00] <jimmy_dean> does gstreamer have any support for ogg theora currently?  I know theora is in alpha, but am just curious
[16:00] <Company> 25% packet loss...
[16:00] Action: LeRoutier is sad. my server died after several hours of DDOS at 4Mb/s yesterday night
[16:00] <Company> jimmy_dean: yup, should just work inside gst-player or totem
[16:00] <jimmy_dean> Company: it works already?  wow
[16:00] <Company> LeRoutier: why would anyone ddos you?
[16:01] <Company> LeRoutier: or did you ddos someone? ;)
[16:01] <LeRoutier> Company, not me personnaly, and not a "real DDOS". seems some machines on my hosting company subnet were infected and trow unlimited garbage to all other machines of the company.
[16:02] <Company> lol
[16:02] <LeRoutier> random source IP addresses (lets say IP spoof), dest = 255.255.255.255 (happy broadcast), length = 40 bytes
[16:02] <Company> that sounds bad
[16:03] <LeRoutier> got 2Gb incoming data in 1h
[16:03] <LeRoutier> that was at 5am, when i was still able to connect to it and when it still replied to ping sometimes
[16:06] <Company> hm
[16:09] <Company> seems the wireless applet is talking crap
[16:09] <Company> the connection seems pretty good for "N/A"
[16:09] <Company> or it's just the airport driver that sucks
[16:10] ensonic ([email protected]) left irc: Read error: 113 (No route to host)
[16:11] <thomasvs> yeah, I have the same problem with my binary wireless module
[16:12] <thomasvs> jonathang: it should work like that, just sounds like you need to recompile the whole stack
[16:12] <jonathang> thomasvs, i'v just recompiled the whole stack!
[16:13] <jonathang> thomasvs, now, i'm recompiling for system wide installation...
[16:22] <thomasvs> jonathang: ok, so grep for where that symbol is being used then
[16:22] <thomasvs> installing it won't do any good
[16:23] <jonathang> in debug_ui i think
[16:23] trow ([email protected]) joined #gstreamer.
[16:23] <jonathang> thomasvs, libs/gst/debug-ui/debug-ui.c
[16:24] <jonathang> and the corresponding header
[16:24] bitshifter ([email protected]) joined #gstreamer.
[16:31] walters ([email protected]) joined #gstreamer.
[16:48] <jonathang> thomasvs, now that i installed everything system-wide, gst-editor is working fine :-s
[16:50] kwm ([email protected]) joined #gstreamer.
[17:15] mxpxpod ([email protected]) joined #gstreamer.
[17:18] ensonic ([email protected]) joined #gstreamer.
[17:19] bitshifter ([email protected]) left irc: "Client exiting"
[17:27] <jdahlin> Tack: http://bugzilla.gnome.org/show_bug.cgi?id=140058
[17:28] <Tack> jdahlin: Cool.  Now, any ideas why that mpeg doesn't play for me?  I tried updating mpeg2dec (to 0.4.0) but that didn't help.  It was a bit of a stab in the dark since I'm not really sure if gstreamer even uses that :)
[17:28] <thomasvs> Tack: does it not play, or does it halt
[17:28] <thomasvs> ?
[17:29] <jdahlin> Tack, sorry, I have no idea, maybe broken registry?
[17:29] <jdahlin> thomasvs: btw, do you know why it's not playing at fullscreen here? it's not resizing correct (with totem) ?
[17:29] <Tack> thomasvs: Neither, exactly.  It seems to be reading the resolution properly and a window pops up, but it remains black.  There's no audio, and I get a lot of:
[17:30] <Tack> [mpeg2video @ 0x409ea648]invalid cbp at 36 11
[17:30] <Tack> [mpeg2video @ 0x409ea648]concealing errors
[17:30] <Tack> [mpeg2video @ 0x409ea648]Warning MVs not available
[17:30] <Tack> [mpeg2video @ 0x409ea648]ac-tex damaged at 5 24
[17:30] <Tack> [mpeg2video @ 0x409ea648]concealing errors
[17:30] <thomasvs> Tack: oh, then it's using the ffmpeg code for mpeg2 playback
[17:31] <jdahlin> I think I found out now
[17:31] <thomasvs> Tack: well, you could try removing the ffmpeg plugin for the test, then it will use the mpeg2dec-based plugin
[17:31] <jdahlin> thomasvs: it should work with ffmpeg too no?
[17:32] <thomasvs> Tack: you may want to revert to a 0.3.2 version though, I believe 0.4.0 removed some symbols
[17:32] <thomasvs> jdahlin: yeah, it should for sure, it's a bug if it doesn't
[17:32] <Tack> thomasvs: Figures.  0.3.2 is what I upgraded from :)
[17:32] <thomasvs> jdahlin: I was just responding to Tack's wanting to use mpeg2dec
[17:33] <Tack> Is there a simple way to remove the plugin?  Will moving libgstffmpeg.so out of the lib/gstreamer directory Just Work?
[17:33] <thomasvs> Tack: if you run gst-register after it, yes
[17:33] <Tack> k ...
[17:34] <Tack> thomasvs: k, it doesn't play at all now :) 
[17:34] <Tack> [tack@viper gstplay]$ ./player.py /home/tack/clip.mpg
[17:34] <Tack> stream length: 00:00:00.000000
[17:34] <Tack> video size 16 16
[17:34] <Tack> [tack@viper gstplay]$
[17:39] <thomasvs> jdahlin: can your example app switch schedulers ?
[17:40] <jdahlin> thomasvs: yes, it parses gst popt options
[17:40] <jdahlin> so --gst-scheduler=basicgthread should work
[17:40] <thomasvs> Tack: can you try again with that option ?
[17:42] <Tack> Should I put the ffmpeg plugin back?
[17:42] <thomasvs> Tack: no, try with that option and the mpeg2dec plugin
[17:43] sxpert_ ([email protected]) joined #gstreamer.
[17:43] <Tack> thomasvs: No change to behaviour or output.
[17:43] <thomasvs> Tack: hmph, weird.  where can I get this clip ?
[17:43] <Tack> sault.org/~tack/clips/
[17:44] <Tack> It's the mpeg.
[17:45] <Tack> Oh, with ffmpeg, it's not completely black if I let it go for a second or two.  There's some resemblance of what is actually video on the screen, though it's badly corrupted, and the audio is coming out in the occasional blip. :)
[17:48] <taaz> clip.mpg played fine with mpeg2dec (the binary app)
[17:48] <taaz> mpeg2dec -s clip.mpg
[17:49] <ensonic> cu
[17:49] ensonic ([email protected]) left #gstreamer ("Leaving").
[17:49] <thomasvs> yep, here too
[17:49] <Tack> Yeah, it does.
[17:49] <thomasvs> damn, I updated to 0.4.0 locally
[17:49] <Tack> For me here as well.
[17:49] <taaz> you really should use 0.4.0 too
[17:49] Action: thomasvs looks for older rpms
[17:49] <taaz> why would you use something that's years old?
[17:49] <thomasvs> years old ?
[17:50] <thomasvs> mpeg2dec 0.3.2 isn't that old
[17:50] Action: taaz heads off to lunch
[17:51] kmaraas ([email protected]) left irc: "Leaving"
[17:54] <jonathang> any ideas why "gst-launch sinesrc ! speed ! osssink" gives me "ERROR: pipeline doesn't want to play." ?
[17:55] <thomasvs> jonathang: probably a bug in speed
[17:55] <thomasvs> jonathang: not negotiating correctly.
[17:56] <jonathang> should i file it in bugzilla?
[18:00] sxpert ([email protected]) left irc: Read error: 110 (Connection timed out)
[18:00] <jonathang> gstpad.c(1246):gst_pad_link_call_link_functions: pad speed0:sink doesn't accept caps
[18:03] mxpxpod ([email protected]) left irc: "Leaving"
[18:04] <jonathang> i admit i never filled a bug before. so if someone is kind enough to help a poor little newbie... :)
[18:04] <thomasvs> jonathang: these days it's simple.  just submit a bug at bugzilla.gnome.org, chose gst-plugins as the component, and put your pipeline and result in the description
[18:05] <Tack> Then the magic GNOME pixies will fix your bug. :)
[18:05] <jonathang> with the debug log?
[18:05] <jonathang> Tack, cool :)
[18:05] <thomasvs> jonathang: possibly, yeah
[18:05] <jonathang> ok
[18:05] <thomasvs> hm, crap, my HEAD has a different bug
[18:07] ChrisHJW ([email protected]) joined #gstreamer.
[18:07] <thomasvs> great.  it's plugging dvddemux :) :)
[18:09] Action: thomasvs fixes
[18:10] <thomasvs> ok, now it works for me
[18:10] <thomasvs> Tack: I'll commit, you can test in a bit
[18:11] <Tack> thomasvs: Cool. :)
[18:13] sub_pop ([email protected]) left irc: "Client exiting"
[18:15] <tlg> is there any crossfader plugin for gst ?
[18:16] sub_pop ([email protected]) joined #gstreamer.
[18:16] Action: jonathang is surprised how submitting a bug report is simple
[18:17] <thomasvs> Tack: ok, try updating
[18:17] <thomasvs> tlg: there's a mixing plugin
[18:17] <thomasvs> Tack: my fix only applies to when not using ffmpeg however, so make sure it's still out of the way
[18:18] <Tack> thomasvs: gst-plugins?
[18:18] <Tack> Oh, gst-ffmpeg
[18:18] <Tack> Oh.
[18:18] <Tack> Misread :)
[18:18] <thomasvs> Tack: no, update gst-plugins and recompile
[18:18] <thomasvs> the problem (for me at least) was the ddvd demuxer being used to decode the mpeg stream
[18:18] <thomasvs> now it's using mpeg2dec, and it works fine
[18:18] <jonathang> tlg, there's an example of how to do sort of a crossfader using volenv and adder in gstreamer/examples/mixer
[18:18] <thomasvs> (against 0.4.0, here)
[18:19] <tlg> thx
[18:19] <tlg> I'll have a look
[18:37] <Tack> thomasvs: Hmmm.
[18:38] <Tack> thomasvs: Well it plays at like 0.1 frames per second and has no audio, but it's not corrupt anymore :)
[18:38] <Tack> It's not even a CPU issue -- the process doesn't even register.
[18:38] <jdahlin> Tack, I think it's about 1fps, if so it's a known issue
[18:39] kmaraas ([email protected]) joined #gstreamer.
[18:39] <Tack> Nope, definitely more like .1 fps.
[18:39] <Tack> Maybe .2
[18:39] <Tack> Yeah, more like .2 or .3.  1 frame every 3-6 seconds.
[18:40] <jdahlin> ok, probably a scheduler bug
[18:40] <jdahlin> can you try with --gst-scheduler=basicgthread on the command line?
[18:41] <Tack> No change.  player.py parses those options?
[18:41] <Tack> Actually, I lied ...
[18:41] <Tack> Now it's closer to 1fps :)
[18:41] Nick change: sxpert_ -> sxpert
[18:41] <Tack> Or is it.
[18:42] <jdahlin> humm :)
[18:42] <LeRoutier> Tack: 1 fps = bad negotiation (it's a fallback)
[18:42] <Tack> Naw, seems about the same.  
[18:42] <Tack> jdahlin: Does player.py parse those args?
[18:42] <jdahlin> Tack: yes, it should
[18:42] <jdahlin> Tack: otherwise, you can try with gst-player
[18:43] <jdahlin> or totem if you have a gnome 2.6 tree installed
[18:43] <Tack> Well except that again I can't control gst-player remotely.
[18:43] <Tack> jdahlin: I'm looking at player.py and it doesn't look like it parses those args.
[18:43] <jdahlin> Tack: it's done when importing gst (uyk!)
[18:43] <Tack> Eww. :)
[18:44] <jdahlin> Tack: so ./player.py --gst-scheduler=basicgthread clip.mpeg should do it
[18:44] <Tack> Yeah, 1fps with that.
[18:44] <jdahlin> ok
[18:44] <Tack> And ... bahh.
[18:45] <jdahlin> ?
[18:45] <Tack> Han on ...
[18:45] tlg_ ([email protected]) joined #gstreamer.
[18:47] <Tack> Ok.  It's difficult to say by visual inspection because the clip is a cartoon, and cartoons generally have fewer unique frames anyway.  So I think it probably is 1fps with either basicgthread or not.
[18:48] <jdahlin> sorry, cannot help you with that then
[18:48] Action: jdahlin looks at Company
[18:48] <thomasvs> Tack: can you try any other mpeg ?
[18:48] <thomasvs> one without cartoon ?
[18:48] <thomasvs> Tack: and, maybe try with the player binary in gst-plugins/examples/gstplay
[18:49] <thomasvs> Tack: also - this is on fc2test2 right? are you using alsa or oss ?
[18:49] md` ([email protected]) left irc: 
[18:49] <Tack> No, RH9, but it's Alsa.
[18:50] <Tack> My RH9 box is the one connected to my TV which I shell into, and has cvs gstreamer.
[18:51] <Tack> Ok, trying non-cartoom ...
[18:51] <thomasvs> hm, so are you using alsasink or osssink ?
[18:52] <Tack> Well, isn't that a good question.
[18:52] <Tack> osssink.
[18:53] <Tack> I changed it to alsasink and now it doesn't play any video.  Sits there black.
[18:53] <Tack> It is indeed about 1fps.
[18:54] BBB ([email protected]) joined #gstreamer.
[18:54] <Tack> gst-plugins/examples/gstplay -- same results.
[18:54] <thomasvs> the weird thing is it ought to work when you change the scheduler
[18:54] <thomasvs> and even weirder is that it works for me
[18:54] <thomasvs> though in the past I had the same issue
[18:54] <BBB> what issue?
[18:54] <BBB> oi
[18:54] Action: thomasvs is sort of pissed we lost mpeg playback along the way
[18:54] <BBB> need to go
[18:54] <BBB> eek
[18:54] Action: BBB kicks himself
[18:54] <thomasvs> BBB: fix fucking mpeg !
[18:55] <Tack> In gconf there's a /system/gstreamer/default and /system/gstreamer/0.8/default -- which one do I change?
[18:55] <BBB> thomasvs: I did!
[18:55] <thomasvs> the 0.8 one.
[18:55] <thomasvs> Tack: the other is from older gst
[18:55] <BBB> what's broken?
[18:55] <thomasvs> BBB: obviously not, since lots of people still have the same issue :)
[18:55] <BBB> the issue of slow playback?
[18:55] <thomasvs> BBB: same as always.  1fps playback
[18:55] <BBB> I don't understand why...
[18:55] <Tack> Ok, because /system/gstreamer/default/audiosink says "osssink sync=false" so I wasn't sure if that was relevant.
[18:55] <BBB> ugh@me
[18:55] <thomasvs> Tack: yeah, old gst
[18:55] <BBB> why does MPEG have that but not AVI and the others?
[18:56] Action: BBB will debug that further tonight
[18:56] <BBB> make a bugreport so I don't forget, assign to me
[18:56] <Tack> You know ...
[18:56] <Tack> With xine on this box I get similar problems.
[18:56] <Tack> Oh hey, it just segfaulted.
[18:57] <Tack> gst-plugins/examples/gstplay
[18:57] <Tack> Lesse ...
[18:57] <thomasvs> BBB: oh, I have better ways of making you not forget :)
[18:57] <jdahlin> BBB, what's your mail?
[18:57] <BBB> [email protected]
[18:58] <BBB> bbl
[18:58] Action: jdahlin reassigns 140058 to BBB
[18:58] <Tack> Here's some interesting behaviour ...
[18:59] <Tack> With this one clip, several frames in, it seems to freeze on one particular frame.  When I use basicgthread, it segfaults.  I'll do a backtrace.
[19:01] <Tack> It's segfaulting in libmpeg2.  Hrm.
[19:02] tlg ([email protected]) left irc: Read error: 110 (Connection timed out)
[19:06] Nick change: tlg_ -> tlg
[19:06] <thomasvs> taaz: offtopic question - why does mpeg2dec fuck up .so versioning on each release ?
[19:08] The_Company ([email protected]) joined #gstreamer.
[19:09] Company ([email protected]) left irc: "Client exiting"
[19:09] The_Company ([email protected]) left irc: Client Quit
[19:09] Company ([email protected]) joined #gstreamer.
[19:11] <Tack> Oh, http://www.netfunny.com/rhf/jokes/01/Apr/mcrhf.html is good. :)
[19:15] <Tack> For those who are doomed to have your media PCs in your living room like I am, I just picked up http://www.systemcooling.com/arcticcooling_cs2l-01.html today and I highly recommend it.
[19:20] BBB ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[19:28] <taaz> thomasvs: you mean why doesn't it use versioning at all?  ask walken.  i've been asking him to version it for years and he doesn't know how or want to or something.
[19:29] <thomasvs> taaz: I hate getting bugs that say stuff like "unresolved symbol mpeg2_pts"
[19:29] <thomasvs> because people upgraded their libmpeg2 rpm
[19:30] <taaz> all i can say is: tough luck.
[19:30] <Tack> Just add Requires: libmpeg2 = 0.3.2 instead of >= ?
[19:30] <taaz> you need to depend on a known package version
[19:30] <Tack> er, mepg2dec
[19:31] <taaz> library deps themselves just wont work since they are unversioned.
[19:31] <thomasvs> or, in other words, walken sucks :)
[19:31] <taaz> yes
[19:32] <taaz> i believe the excuse in the past was that he didn't want to commit to an interface enough to version it.  which of course makes no sense at all.  i may have misunderstood what he was saying about that though.
[19:33] <taaz> send a mail to the libmpeg2 dev list pleading for versioning along with a patch with included doc file for updating versioning procedures and you may get a positive response
[19:34] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[19:45] <LeRoutier> bbl
[19:45] LeRoutier ([email protected]) left irc: "Leaving"
[19:50] walters ([email protected]) left irc: Client Quit
[20:08] pb_ (~pb@2002:3e03:42cc:1:20b:6aff:fe19:f290) left irc: "Client exiting"
[20:10] thomasvs ([email protected]) left irc: Read error: 113 (No route to host)
[20:30] thomasvs ([email protected]) joined #gstreamer.
[20:34] md` ([email protected]) joined #gstreamer.
[20:34] jdahlin ([email protected]) left irc: "Leaving"
[20:35] dolphy ([email protected]) joined #gstreamer.
[20:36] Nick change: bilboed_ -> bilboed
[20:51] md` ([email protected]) left irc: 
[20:53] wheels ([email protected]) joined #gstreamer.
[21:03] pb_ (~pb@2002:5160:4550:1:240:95ff:fe30:aa84) joined #gstreamer.
[21:04] <Company> it's refreshing to have a build that just works
[21:04] Action: Company cheers for thoimasvs while kicking libgda
[21:04] Action: Company cheers for thomasvs, too
[21:07] md` ([email protected]) joined #gstreamer.
[21:20] BBB ([email protected]) joined #gstreamer.
[21:21] <BBB> thomasvs: I might do a commit tonight, depending on whether I get my thing working... (ffmpeg playback thingy)
[21:25] jdahlin ([email protected]) joined #gstreamer.
[21:26] <BBB> hey dudes
[21:26] <BBB> ted mailed me to re-ask me to take over maintainership over gnome-media... should I do that?
[21:26] <jdahlin> interested?
[21:26] md` ([email protected]) left irc: 
[21:27] jonathang ([email protected]) left irc: "Leaving"
[21:27] <jdahlin> you'd have to do releases and stuff :)
[21:27] <BBB> me and releases? oh my, poor world ;)
[21:27] <jdahlin> time to grow up? :)
[21:27] <BBB> a mjpegtools release generally takes several months, not to mention a gst-rec release
[21:28] <BBB> which has taken me more than two years now
[21:28] <BBB> and still not even close
[21:28] Action: BBB is basiaclly thinking about it
[21:28] Action: wheels generally only releases when it becomes eaiser than dealing with the continued bitching for him to do a release.  ;-)
[21:28] <BBB> not sure if I'm the right person for the job, it'd be easy to bring gstreamer into gnome-cd and related tools, though
[21:29] <BBB> and I actually want totem in gnome-media as soon as the gst-version is "good" enough
[21:29] <BBB> not sure if hadess agrees
[21:29] <jdahlin> is totem in gnome yet?
[21:29] <BBB> no, but this'd be the way to get it in ;)
[21:29] <jdahlin> BBB, I'd say take it
[21:29] <jdahlin> doing releases is not that hard
[21:29] <BBB> put a gst-based totem in gnome-media when it's grown-up
[21:30] <jdahlin> just make sure distcheck pass and the number of open bugs are low
[21:30] <ds-work> bump the version, make distcheck
[21:30] <BBB> nobody wants to download yet another 50 extra packages as part of the D&DP
[21:30] <jdahlin> BBB, it's another way of "infiltrate gnome"
[21:30] <BBB> I don't think gnome-media has that many open bugs
[21:30] <BBB> I ususally close all the new ones as dups of old ones ;)
[21:30] <jdahlin> and make sure gstreamer stays there :)
[21:30] <BBB> or I re-assign to totem
[21:30] <BBB> heh :)
[21:31] <jdahlin> are you still mad on me cause of all "style changes" in grecord?
[21:31] <ds-work> BBB: btw, did you notice that caps now fully supports buffers?
[21:31] <BBB> I don't think I was ever mad... just hated it because I had some local changes that got lost there ;)
[21:31] <BBB> ds-work: I saw it in CVS, I didn't test it out yet
[21:31] <BBB> I will soon
[21:31] <ds-work> audio/x-whatever,field=(buffer)1a2b3c4d
[21:32] <jdahlin> BBB, d'oh!
[21:32] <BBB> (buffer)0xhex?or (buffer)"somestring"?
[21:32] <BBB> jdahlin: GSR works, no? :)
[21:32] <BBB> jdahlin: that was the most important thing
[21:32] <jdahlin> BBB, it works in most cases now :)
[21:33] <BBB> good enough ;0
[21:33] <BBB> flac encoding still fails locally
[21:33] <jdahlin> Uraeus broke it
[21:33] <BBB> I need to look into that some day
[21:33] <BBB> he did?
[21:33] <BBB> oi
[21:33] <jdahlin> but that's only norweigans
[21:33] <jdahlin> :)
[21:33] <BBB> kick him
[21:33] <jdahlin> I will, I will
[21:34] <BBB> good ;)
[21:34] <BBB> ~feel the love tonight~
[21:34] <jdahlin> BBB, did you look at the msmpeg4/ffmpeg bug I submitted?
[21:34] <BBB> I saw it, didn't look into it yet
[21:34] <BBB> will do so soon
[21:34] <BBB> I hope I'll be able to fix it before 0.8.1
[21:34] <jdahlin> I wonder how it works in mplayer
[21:34] <BBB> don't count on it, though
[21:35] <jdahlin> Tack does :)
[21:35] <wheels> Is there a planned release schedule for 0.8.1 yet?
[21:35] <BBB> maybe extra data in the stream header
[21:35] <BBB> like for msmpeg4v2
[21:35] <BBB> some files decode even without it
[21:35] <BBB> others don't
[21:35] <Tack> Well, it'd be nice, but it's not like my life depends on it. :)
[21:35] <BBB> I have local changes to use this stream header data
[21:35] <BBB> but... nothing final yet
[21:36] <BBB> I'm having troubles getting palettized data to work right now, other magic stuff in ffmpeg
[21:36] <jdahlin> BBB, the funny part is that mplayers msmpeg4.c is exactly the same as the one in gstreamer now...
[21:36] <BBB> wheels: thursday,hopefully
[21:36] <wheels> ah
[21:36] <wheels> hmm
[21:36] <jdahlin> probably something at plugin/element level
[21:36] <BBB> jdahlin: they have more experience setting up the AVCodecContext
[21:36] <BBB> I still need to learn a lot
[21:37] <BBB> but I only do so at a per-case level
[21:37] <BBB> there's no documentation on it :(
[21:37] <jdahlin> BBB, why not copy the mplayer code?
[21:37] <BBB> GPL... :(
[21:37] <jdahlin> copy ideas?
[21:38] <jdahlin> ask them to relicence parts of it?
[21:38] mxpxpod ([email protected]) joined #gstreamer.
[21:39] <BBB> jdahlin: heh... these people *hate* the LGPL
[21:39] <jdahlin> BBB, really?
[21:39] <BBB> they see any sort of corporate involvement, like we're doing, as pure evilry conspiracy to take over the opensource world and so on
[21:39] <BBB> these people are completely nuts
[21:39] <BBB> michael niedermayer is ok
[21:39] <jdahlin> teenagers
[21:39] <wheels> BBB: Well, it is, right?
[21:39] <wheels> BBB: Evil, that is.
[21:40] <BBB> sure is
[21:40] <thomasvs> Company: heh, thanks :)
[21:40] <BBB> because we'll get better and better and they'll continue sucking
[21:40] <BBB> I read on the ffmpeg-devel list that the next redesign of g2 is delayed until june
[21:40] <BBB> so we have some more time now ;)
[21:41] <thomasvs> BBB: if you don't want gnome-media, or want to share with one of us, that's ok too
[21:41] <BBB> I'm happy enough that ffmpeg is LGPL, though
[21:41] <BBB> wanna take it together thomasvs? you can do releases, I'll do patches, bugzilla and so on
[21:41] <BBB> I've been doing that for a while anyway
[21:41] <jdahlin> is xine lgpl?
[21:41] Zeenix ([email protected]) joined #gstreamer.
[21:42] <Zeenix> hi
[21:42] <Tack> Next redesign of g2 as in mplayer g2?
[21:42] <BBB> tack: yes
[21:42] <BBB> jdahlin: no
[21:42] <BBB> jdahlin: I ususally take ideas straight from ffmpeg's demuxers
[21:42] <Tack> g2 is tantalizing vapour. :)
[21:42] <BBB> but those are fairly cryptic and uncommented
[21:43] <BBB> tack: yup ;)
[21:44] <BBB> jdahlin: wanna take a piece of gnome-media maintainance as well? ;)
[21:44] <jdahlin> BBB, not just yet, ask the belgian guy
[21:44] <BBB> it means that I need to keep up with unstable gnome versions though
[21:44] Action: BBB hates that
[21:44] <jdahlin> BBB, just wait a couple of weeks for FC2
[21:44] <jdahlin> then you'll get pretty much the latest
[21:45] mxpxpod ([email protected]) left irc: "Ex-Chat"
[21:45] <BBB> I'll try
[21:46] <wheels> BBB: I'm sure Gentoo has some ebuilds for it.  :-)
[21:46] <BBB> nah :-p
[21:46] mxpxpod ([email protected]) joined #gstreamer.
[21:47] <lilo_booter> well, i'm hoping that we've finalised this release of mlt (if anyone cares to test it, i'd be grateful :-)) - think i got enough info from thomasvs this morning to provide a gst demuxer for our xml format... :-)
[21:47] <wheels> lilo_booter: any update on the license thing?
[21:48] <lilo_booter> wheels: ah - forgot to ask - will be talking to him again tomorrow or the day after i think
[21:48] Action: BBB pokes thomasvs
[21:49] <thomasvs> BBB: yeah, sure
[21:49] <thomasvs> BBB: put me up for comaintainer
[21:51] <thomasvs> BBB: I did a bunch of the last releases anyway, and I've been wanting to clean up the build some more
[21:55] <ds-work> what's wrong with vmsize()?
[21:55] <thomasvs> it doesn't seem to work correctly
[21:56] <thomasvs> the tests give different results over multiple runs
[21:56] <thomasvs> check some of buildbot's logs for rh9 and ydl
[21:56] <ds-work> ok
[21:56] <ds-work> we should reenable the tests, but not use the results of vmsize() for the exit status
[21:57] <thomasvs> maybe; they don't say much then though do they ?
[21:57] <ds-work> if we've got code, we might as well run it in the testsuite
[21:57] <ds-work> it should at least run to completion
[21:57] <thomasvs> sure, but it ought to test stuff
[21:57] <thomasvs> (and probably test stuff we think is valid behaviour :))
[21:58] <ds-work> in that case, we might as well disable most of testsuite/
[22:02] <thomasvs> I don't know.
[22:02] <thomasvs> why are you asking me anyway ? :)
[22:15] Shoragan ([email protected]) left irc: "Leaving"
[22:19] Action: Zeenix pretty sure thomasvs is finished with the vw-howto..
[22:21] <BBB> is there something like gstreamer-cvs for gnome cvs?
[22:21] <BBB> so notification of changes?
[22:21] <Company> BBB: there's one big list
[22:22] <BBB> for the whole of gnome?
[22:22] <BBB> oh my
[22:22] <Company> http://mail.gnome.org/mailman/listinfo/cvs-commits-list
[22:22] <Company> you're probably supposed to filter it
[22:23] Action: BBB only filters in evolution
[22:23] Action: Company doesn'T filter at all
[22:24] <Company> that's why i'm only subscribed to 4 lists - 3 low post count and gst-devel ;)
[22:25] Action: wheels has cut back to 15 lists.  :-)
[22:25] Action: Zeenix was FORCED to learn filtering by crazy kernel guys
[22:25] <Company> filtering doesn't work because i (unlike other people) rely on my mail staying on the server
[22:26] <Company> and i have no access to my uni mail setup
[22:26] <thomasvs> Company: procmail ?
[22:26] <thomasvs> BBB: did we agree to agree silently now, or ... ?
[22:27] Action: ds-work has a 200-line .procmailrc
[22:28] <Company> jdahlin: xine is GPL
[22:28] <BBB> thomasvs: yes
[22:28] Action: Company is reading backlog
[22:28] <BBB> oh, right, so in case you didn't notice,I wasn't really here
[22:28] <Company> gstreamer is the only lgpl media framework
[22:28] <BBB> Zeenix: I gave up linux-kernel after a while
[22:28] <BBB> impossible to keep up with
[22:29] <BBB> and most bug reports go through private mail anyway
[22:29] <jdahlin> Company: even xine-libs?
[22:29] <BBB> xine-libs is GPL too
[22:29] <Company> jdahlin: yes
[22:29] <BBB> we're not based on avifile or anything
[22:29] <jdahlin> that makes gstreamer the choice for any corporations
[22:29] <BBB> that's our strong point
[22:29] <BBB> ;)
[22:29] <Company> yes
[22:29] <jdahlin> it's just a matter of time before we pass mplayer/xine
[22:30] <Company> and yes again
[22:30] <BBB> our weak point si that mplayer can steal any code from us
[22:30] <jdahlin> doesn't matter
[22:30] <BBB> they're currently apparently "developing" a ffmpeg demuxer wrapper
[22:30] <Tack> BBB: Is that really a weak point?
[22:30] <Company> yeah, you don't win by copying code verbatim
[22:30] <jdahlin> as soon as we get stable and application starts to use us
[22:30] <BBB> like I did one month ago in gst-ffmpeg
[22:30] <jdahlin> then nobody wants to use mplayer
[22:30] <BBB> I think they're just copying mine, like usually ;)
[22:30] <BBB> tack: dunno, maybe
[22:31] <BBB> mplayer is still better than us
[22:31] <BBB> so is xine
[22:31] <BBB> we're getting closer, though
[22:31] <Tack> BBB: Vastly so. :)
[22:31] <Company> mplayers ui sucks
[22:31] <jdahlin> Tack, I wouldn't say vastly
[22:31] <Company> xine's ui doesn't though ;)
[22:31] <Tack> mplayer continues to impress me in that it plays whatever I throw at it.
[22:31] Action: BBB thinks that a good media playback support thing will be enough to get totem-gst in gnome
[22:31] <Tack> But yes, I've seen its code, and it's scary. :)
[22:31] <jdahlin> There are more important things then just playing a simple movie from disk
[22:31] <Company> i had mplayer segfault for avis on ppc
[22:31] <Company> and i had it not compiling
[22:31] <Tack> jdahlin: There are, agreed, but 99% of most users don't care and just want a media player That Works.
[22:32] <Company> and apt-get install mplayer still doesn't work
[22:32] <jdahlin> Company: is ffmpeg of any kind in debian?
[22:32] <BBB> that's always a nice topic yes ;)
[22:32] <Company> though mplayer is still the gold standard
[22:32] <jdahlin> Tack, well, some users want a nice ui, like totem
[22:32] <BBB> jdahlin: notanymore,I think
[22:32] <BBB> not in default
[22:32] <Company> jdahlin: yes, part of xine and part of gstreamer
[22:32] <BBB> oh, it is?
[22:32] <Tack> jdahlin: Yes, well true, but nice UI is the secondary concern.
[22:32] <BBB> that's surprising then :)
[22:33] <Company> yeah, it is
[22:33] <Company> there's no "ffmpeg" in there
[22:33] <jdahlin> Tack, for the current users, not for the future ones
[22:33] <Company> but there's gstreamere-ffmpeg and xine
[22:33] <jdahlin> Tack, we don't want to build something only powerusers can use
[22:33] <jdahlin> make it easy dammit!
[22:33] <BBB> totem is easy :)
[22:34] <Tack> jdahlin: I quite disagree.  A friendly interface is important, but rather irrelevant if the pretty, usable interface can't play 50% of a collection, and crashes on 25% of the rest of it.
[22:34] <BBB> and to get into GNOME, you need a damn good UI
[22:34] <jdahlin> I know all keyboard shortcuts in mplayer by heart, and I'm very scared
[22:34] <Tack> Definitely agree that a good UI is necessary and important, but it's still secondary.
[22:34] walters ([email protected]) joined #gstreamer.
[22:34] <Tack> What I'm saying is that gstreamer is more important than totem. :)
[22:34] <jdahlin> both are primary goals
[22:34] <jdahlin> but making a good ui is kinda trivial
[22:35] <jdahlin> compared to making all media files work :)
[22:35] <Tack> Yeah. :)
[22:35] <Company> nope
[22:35] <Tack> And making a good ui isn't trivial.
[22:35] tlg ([email protected]) left irc: "Client exiting"
[22:35] <Company> totem and gstreamer are equally important
[22:35] <Company> there's a huge differewnce between having an app that uses your code and not having one
[22:35] <jdahlin> Tack, not exactly trivial but a lot easier than testing 1000s of video files and making sure they work
[22:35] <jdahlin> and not breaking cvs all the time!
[22:36] <Tack> It's easy for me to believe regressions are a serious problem for a project like gstreamer.
[22:36] <Company> there are no regressions
[22:36] <Tack> Company: Yeah, but the first step is making sure the library is solid.  
[22:36] <BBB> well, we're getting better and better
[22:36] <jdahlin> Tack, you don't wanna know
[22:36] <jdahlin> but we have a plan
[22:36] <Company> if there were regressions, people would supply tests that worked and don't anymore and our testsuite would grow ;)
[22:36] Action: BBB is working on media playback improvements all day
[22:37] <Company> Tack: a library only becomes solid by testing
[22:37] <jdahlin> to have a large testsuite, that tests all media files on a regular basis
[22:37] <Tack> Company: Chicken and egg, sure.
[22:37] <Company> Tack: for testing you need users
[22:37] <jdahlin> and that will annoy the developer who broke it till death if he's not fixing it
[22:37] <Company> Tack: gstreamer's media handling sucked (and partially still sucks) because noone finds the weird files to throw at it
[22:38] <jdahlin> Company: just wait, I'm going to throw the entire samples directory in mplayer at it
[22:38] <Company> jdahlin: please
[22:38] <Company> jdahlin: and don't forget seeking ;)
[22:38] <jdahlin> Company: but we need to have the testing framework first
[22:38] <jdahlin> Company: I think seeking is secondary
[22:38] <Company> jdahlin: yeah, and a testfiles repository
[22:39] <jdahlin> Company: imporant, but simple playback is more important
[22:39] <jdahlin> important
[22:39] <Company> well....
[22:39] <BBB> seeking is important thouh
[22:39] <Company> the problem is that gst-launch always worked perfectly
[22:39] <jdahlin> we need a way to do seeking on the command line
[22:39] <BBB> if you can't seek, well...
[22:39] <BBB> jdahlin: navseek! :)
[22:39] <Company> you can still see that in gst
[22:39] <jdahlin> BBB, does that work?
[22:39] <BBB> yes
[22:39] <BBB> with arrows even
[22:39] <Company> like ther's no way to disconnect elements and state changes only work once
[22:40] <Tack> Company: You want tests to throw at it?  Go to some alt.binaries.erotica group.  If gstreamer can handle all the formats and broken files in there, that's worth bragging about :)
[22:40] <Company> stuff like that
[22:40] <jdahlin> like, seeking to 10th second of file
[22:40] <Tack> Hmm, indeed.  Does/will gstreamer support seeking to non-key frames?
[22:40] <Company> it should
[22:41] <Company> there's a GST_SEEK_CERTAIN flag
[22:41] <Company> though that can't be done inside demuxers
[22:41] <Company> it requires help from the decoder elements
[22:41] <Tack> Yeah.
[22:41] <Tack> But it'd be a really nice feature :)
[22:41] <Company> someone needs to code it ;)
[22:42] <ds-work> decoders need to take a seek and turn it into a keyframe seek
[22:42] <ds-work> then drop the frames until the correct one
[22:42] <BBB> yup...
[22:42] <BBB> it's hard though
[22:42] <ds-work> yeah, that's really hard
[22:42] <Tack> ds-work: Almost. 
[22:42] <Tack> Well, no, nevermind, that's it. :)
[22:42] <Company> and demuxers need to decide to seek to the keyframe before the requested seek
[22:43] <Company> avi or mpeg seeks to nearest
[22:43] <Tack> You take a position, find the closest key frame behind that position, and then decode, dropping frames until you hit pos.
[22:43] <jdahlin> hmm, it should be possible to create an element that just drops the first n bytes of the input stream
[22:44] <ds-work> one could write a seek-fixer element that does the same thing, really
[22:44] <Company> don't we already have that?
[22:44] <Company> no we don't
[22:44] <BBB> anyway, this seeking to exact times is not important right now
[22:44] <Company> breakmydata only modifies it
[22:44] <BBB> seeking should "generally" work
[22:44] <BBB> allthis finetuning and perfectizing can come later
[22:44] <Company> seeking to exact times is important the moment people start video editing
[22:44] <Tack> As long as the architecture allows something like that to be added later without major infrastructure overhausl :)
[22:44] <BBB> most people won't even notice or care
[22:44] <BBB> Company: first video playback, then video editing
[22:45] <BBB> oh, and I have an exact-seek patch for MPEG
[22:45] <BBB> in 24h files, it's 1-second accurate :)
[22:45] Action: BBB thinks that's pretty impressive
[22:45] <jdahlin> who has 24h mpeg files?
[22:45] <BBB> my company
[22:45] <Tack> jdahlin: I could make one easily :)
[22:45] <BBB> :p
[22:45] <BBB> we have loads of them
[22:45] <jdahlin> Tack, I don't want to copy that
[22:45] <Tack> (cat /dev/video > longfile.mpg); sleep $((60*60*24)); killall cat :)
[22:46] <Tack> Missing a &, but you get the idea.
[22:46] <Company> someone kills kittens here!
[22:46] Action: BBB kicks tack
[22:46] <BBB> foei!
[22:47] <Tack> I'm so ashamed :)
[22:47] <ds-work> gst-launch videotestsrc ! ffenc_mpeg ! filesink
[22:49] <BBB> does that crash?
[22:49] <BBB> if not: wow ;)
[22:49] Action: BBB didn't test ffenc_* for ages now
[22:50] <BBB> try sinesrc ! ffenc_ac3 ! a52dec ! osssink
[22:50] <Company> hm
[22:51] <Company> i should write an audiosink base class soon
[22:51] <BBB> last time I tried, ffenc_ac3 ate lots and lots of memory
[22:51] <jdahlin> Company: interface?
[22:51] <Company> jdahlin: no, base class
[22:51] <BBB> no, base class
[22:51] <BBB> bleh
[22:51] <Company> so there's just 5 or so functions that the backend needs to implement
[22:51] <jdahlin> oh, cool
[22:51] <Company> and all the other stuff that works the same everywhere isn't copied
[22:51] <jdahlin> same for a sourcE?
[22:52] <Company> not now ;)
[22:52] <jdahlin> blargh :)
[22:52] <Company> i'm not that interested in sources
[22:52] <Company> especially because alsalib always crashes when i test them
[22:52] <jdahlin> fix it?
[22:53] <Zeenix> BBB: well, i need it for my job
[22:53] <Zeenix> BBB: sorry for the late reply
[22:53] <Tack>  /home/tack/src/gst-ffmpeg/gst-libs/ext/ffmpeg/libavformat/allformats.c:115: undefined reference to `sgi_image_format'
[22:53] <jdahlin> Tack: cvs up?
[22:53] <Tack> jdahlin: Did just now.
[22:54] <jdahlin> hmm
[22:54] <jdahlin> thomasvs?
[22:54] <Company> jdahlin: i'm not gonna fix alsalib ;)
[22:54] <jdahlin> Tack: try a make clean?
[22:54] <Tack> Indeed.
[22:54] <jdahlin> Tack: compiles for me
[22:54] Action: Tack tries again.
[22:55] <jdahlin> Tack: do you have sgi_image_format in avformat.h ?
[22:55] <taaz> eek.  i need to get these debs done before 0.8.1 comes out ;)
[22:55] <taaz> i think they are done really
[22:56] <Tack> jdahlin: Yes, extern AVImageFormat sgi_image_format;
[22:56] <Tack> but it's extern.
[22:56] <taaz> wheels: got ideas on arts debian problem?
[22:56] <jdahlin> Tack: is this linktime or compile time?
[22:56] <Tack> Oh it's in sgi.c...
[22:56] <wheels> taaz: Haven't looked at it -- but I will in a couple of minutes...
[22:57] Action: wheels finishes up some pending code...
[22:57] <Tack> jdahlin: Compile time I think, but I'm retrying after make clean.
[22:57] <jdahlin> Tack: if it still doesn't work, remove your tree and recheckout
[22:57] <jdahlin> otherwise, bug thomas
[22:57] Action: jdahlin is off to prepare supper
[22:58] <Tack> Oh yes, I'm still getting "gstffmpeg.c:27:21: gst/gst.h: No such file or directory" due to GST_CFLAGS not getting put in the Makefile.
[23:02] <ds-work> could people look at http://www.schleef.org/~ds/people and suggest corrections?
[23:03] <Tack> Still not able to play my clip.avi -- thought I'd try. :)  But, playing matrix.avi, when I use alsasink, I get extremely choppy audio (like it plays for .5 seconds and cuts out for .5 seconds) although the video is smooth.  osssink seems to work fine.
[23:04] <dolphy> ds-work: did i do something bad to be out of that list ? :-D
[23:05] <Company> dolphy: you weren't loud enough the last week ;)
[23:05] <Company> ds-work: you could be a bit more explicit about what the people do ;)
[23:06] <ds-work> dolphy: you weren't in my log for the ~30 hours that I checked
[23:06] Action: dolphy sings a little then. tralalala
[23:07] <ds-work> Company: it's mainly for my benefit so I can remember which people are writing which apps :)
[23:07] <Company> ds-work: ah, ok
[23:07] <ds-work> but it might as well be general purpose
[23:07] <Company> ds-work: bilboed is gnonlin hacker, chrishjw is matroska guy, teuf is rb hacker
[23:08] ploum ([email protected]) left irc: "Leaving"
[23:08] <Company> and he's named Christophe Fergeau
[23:08] <Company> i like that name because it's so complicated ;)
[23:09] <walters> heh
[23:11] <Company> walters: atm i'm fighting gda and its sqlite backend btw
[23:11] <walters> Company: i would have figured there'd be more problems with rhythmdb :)
[23:12] <walters> Company: btw, are you using arch?
[23:12] <Company> walters: no, rhythmdb is just very slow, but at least it works
[23:12] <Company> walters: i don'T have webspace
[23:12] <walters> Company: i can provide you with that if necessary
[23:12] <taaz> ds-work: you could jdahlin to gst-python list since he's doing all the work these days ;)
[23:12] <walters> Company: you also don't actually need web space
[23:12] <Company> there was never a need for me to have some until arch showed up ;)
[23:13] <walters> Company: you can have an archive stored on your local machine, hack away there, and merge to the main archive.
[23:13] <walters> Company: or just use the main archive directly
[23:13] <walters> Company: basically what i'd like is if you could merge "generic" parts of your work right away into --0.9
[23:13] <ChrisHJW> ANNOUNCEMENT :
[23:13] <BBB> ds-work: [email protected] ;)
[23:14] <walters> Company: so if like you did the work for making RhythmDBEntry a struct, that could be merged now
[23:14] <ChrisHJW> The Corecode and matroska people announce a fork of gstreamer, called 'CoreStreamer'
[23:14] Action: ChrisHJW waits to get slapped now
[23:14] <walters> Company: you can sftp [email protected] right?
[23:15] <BBB> ds-work: and I do gnome-media also
[23:15] <Company> walters: yeah
[23:15] <BBB> ChrisHJW: j/k??
[23:15] <ChrisHJW> BBB : nope
[23:15] <walters> Company: ok.  all you need to do then is set up GPG signing of commits and you're good to go
[23:16] Action: ds-work wonders why people would fork projects that they don't contribute to
[23:16] <ChrisHJW> ds-work : its the only chance we see to use Gstreamer
[23:16] ploum ([email protected]) joined #gstreamer.
[23:16] <ploum> Hello
[23:17] <ploum> I cannot play MP3 with gstreamer 0.8 on Debian experimental
[23:17] <ploum> Anyone have the bug ?
[23:17] <ChrisHJW> if we make the Intel compiler mandatory, we will see very little contribution from the Windows world
[23:17] <ChrisHJW> same for GCC
[23:17] <ploum> with gstreamer I cannot play anything , but with 0.8 ogg files are perfectly read
[23:17] <Company> ChrisHJW: i'm not sure why we can't stay in the same cvs repo, especially considering how modular gst is
[23:17] <walters> Company: basically just do: echo "#!/bin/sh"; echo gpg --clearsign > ~/.arch-params/signing/[email protected]
[23:17] <ds-work> ChrisHJW: why does that require forking?
[23:18] <walters> Company: er, oops
[23:18] <walters> Company: (echo "#!/bin/sh"; echo gpg --clearsign) > ~/.arch-params/signing/[email protected]
[23:18] <ChrisHJW> Company : consider my situation, i have no clue about all of this, i see you guys and ym guys discussing back and forth about crazy macros and stuff
[23:18] <Company> ChrisHJW: it's basically just the boring work to get it to compile
[23:19] <ChrisHJW> ds-work : what alternatives do we have if we are not allowed to put our stuff into your CVS ?
[23:19] <Company> ChrisHJW: and when using MSVC it's a lot of boring work ;)
[23:19] <ds-work> btw, HEAD gstreamer + a few patches compiles under MinGW
[23:19] <BBB> ChrisHJW: I proposed a fix 10 times
[23:19] <Company> ChrisHJW: and it's of course figuring out how to do stuff correctly
[23:19] <BBB> but steve dopesn't read that part of my emails
[23:19] <ChrisHJW> Company : i understand my guys thats its not only a lot of work, but for some plugins requires a rewrite
[23:19] <BBB> I told him ten times to just simply add vararg functions for the debug macros in a win32 specific header file
[23:20] <BBB> and the function generating macros can be removed from the code (not headers!!)
[23:20] <BBB> that's all to port it
[23:20] <BBB> but steve and jory don't read those parts of my emails
[23:20] kwm ([email protected]) left irc: "Leaving"
[23:21] <BBB> they're too much blinded by the fact that I don't care about MS or MSVC; it's apparently hard to believe for a win32 user that there's users that really don't care a single bit about win32
[23:21] <ChrisHJW> i dont care about win32
[23:21] <ChrisHJW> i care about matroska
[23:21] <ChrisHJW> and i have to care to make its stuff available for the most used platforms
[23:21] <Company> ChrisHJW: i wasn't talking about plugins, i was just talking about the core
[23:22] <ChrisHJW> Company : ok
[23:22] <Company> ChrisHJW: it's probably a good idea to have your own windows specific plugins
[23:22] smoke ([email protected]) got netsplit.
[23:22] <ChrisHJW> BBB : i copied and pasted your comments to #matroska, and Mosu agrees with you :)
[23:24] <Company> and it's probably important if we want to stay in sync for the next releases
[23:26] <ChrisHJW> Company : our idea was to make the fork only based on your stable versions, and care mostly about working win32 and macosx plugins during that time
[23:26] <ChrisHJW> then, if you proceed to the next stable version, we make the necessary API changes
[23:26] <ChrisHJW> thats how i understood it, at last
[23:27] <jdahlin> ds-work: you could add me to gnome-media and gst-media-test too
[23:27] <Company> ChrisHJW: that's fine unless you want to influence the direction in which the core develops
[23:27] <ChrisHJW> Company : we trust you guys fully on that i guess :) .... you are pro's
[23:28] smoke ([email protected]) got lost in the net-split.
[23:30] <Company> that's your choice :)
[23:30] <ChrisHJW> lets see how it develops
[23:30] <Company> though i'd still hope to keep the core in sync
[23:30] <ChrisHJW> the goal is to be able to launch some first working plugins pretty soon, so we can test the reaction from the windows world
[23:31] <ds-work> does anyone have gstreamer actually working on windows?
[23:31] <ChrisHJW> if they accept the idea of a new framework
[23:31] <ChrisHJW> yes
[23:31] <ChrisHJW> since mor than 4 weeks now
[23:31] <ChrisHJW> we can encode MP3's :)
[23:31] <ds-work> how did they build plugin DLLs?
[23:31] <ds-work> and why don't we get patches?
[23:32] <Company> they didn't have that linking problem yet ;)
[23:32] <Company> they only decode mp3s
[23:32] <ChrisHJW> jcsston ported some plugins over, i guess he has MP3 en/decoding, DivX decoding, AVI parsing working
[23:33] <jdahlin> <luis> NB: BUGZILLA IS GOING DOWN NOW FOR SOME MAINTENANCE
[23:33] <jdahlin> just so you know :)
[23:34] <Company> thx jdahlin
[23:34] <Company> (especially because i wanted to start filing libgda bugs soon )
[23:34] <walters> Company: when you get done battling let me know because i'm really interested in your rhythmdb work :)
[23:36] <BBB> ds-work: I've been following it, don't worry
[23:36] <BBB> ds-work: I told them to hold off patches
[23:36] <BBB> since they use custom vararg macros for now
[23:36] <BBB> I told them that'll never go in
[23:37] <BBB> read: #define GST_DEBUG_1(x) ..
[23:37] <BBB> #define GST_DEBUG_2(x,y) ...
[23:37] <BBB> etc.
[23:37] <BBB> that sort of ...&$
[23:37] <Company> hm
[23:37] <Tack> Ick.
[23:37] <BBB> I told them to use inline vararg functions for now
[23:37] <Company> static inline void GST_DEBUG(...) { }; sounds easier
[23:37] <BBB> but somehow, it takes them a while to understand what I'm saying
[23:37] <BBB> Company: exactly
[23:38] <Company> BBB: you like to pick on each other, eh? ;)
[23:38] <BBB> :)
[23:38] <Company> ds-work could probably even write a perl script that converts that
[23:38] <BBB> read the matroska-devel archive, it's far worse than this :p
[23:38] <Tack> Does MSVC not support vararg macros or something?
[23:39] <BBB> ds-work: and spyder842 is from matroska aswell
[23:39] <ChrisHJW> nope, its not C99 compliant
[23:39] <Tack> Unfortunate.
[23:39] <Tack> They're terribly convenient :)
[23:39] <ChrisHJW> Intel compiler is, but its buggy and slow
[23:39] <ds-work> does MSVC support static inline?
[23:39] <BBB> it does
[23:39] <jdahlin> ChrisHJW, is intels compiler really slow? I thought it was faster than gcc
[23:40] Action: BBB played with MSVC at a previous life's vacation job
[23:40] <ds-work> or rather, _forced_ inlining
[23:40] <BBB> oh, dunno... I'm not a MSVC expert
[23:40] <jdahlin> google can probably tell
[23:40] <BBB> I was rather proud that I got libpng to compile under visual studio
[23:40] <jdahlin> BBB, so what about gstreamer? :)
[23:41] <BBB> nah... I don't own visual studio anymore
[23:41] <BBB> neither do I own a win32 workstation
[23:41] <BBB> someone wanna buy me a MacOS X one? I'll maintain a Mac OS X port ;)
[23:41] <Tack> glib basically just does #ifdef G_HAVE_GNUC_VARARGS) \ ... \ #else \ static void ... \ #endif
[23:42] <ds-work> gstreamer-0.6 is in Fink.  it's being updated to 0.8 currently
[23:42] <BBB> ds-work: and don't forget to mention markey
[23:42] <markey> hm?
[23:42] Action: taaz pokes wheels 
[23:42] <BBB> good for our KDE integration stuff :p
[23:42] <markey> sup?
[23:43] <ChrisHJW> jdahlin : no idea about GCC compared to Intel, sorry
[23:43] ploum ([email protected]) left irc: "Leaving"
[23:43] <jdahlin> ChrisHJW, so what did you compare to? MSVC?
[23:43] Action: ChrisHJW never went beyond creating a test project called 'christian' in MSVC6 trial version ;) ....
[23:44] <ChrisHJW> jdahlin : yes, the guys all say Intel is much slower than MSVC, and buggy
[23:44] <jdahlin> ChrisHJW, and it kicks gcc ass in performance
[23:45] <thomasvs> ds-work: ensonic is Stefan Kost
[23:47] <thomasvs> ChrisHJW: I don't understand - why are we not seeing patches from you guys ? Why do you need to fork to be able to add a few plugins ?
[23:47] <thomasvs> ChrisHJW: seems to me like a fork for no reason
[23:47] <taaz> ds-work: wingo is really working on gst-guile not gst-scheme ;)
[23:47] <thomasvs> ChrisHJW: and I've seen zero on this on our devel list, what's going on ?
[23:47] <ChrisHJW> thomasvs : are you reading our list ? there was plenty of it, at least 30 mails recently
[23:48] <ChrisHJW> oh, you mean your list
[23:48] <thomasvs> ChrisHJW: duh, yeah
[23:48] <ChrisHJW> thomasvs : dont worry, the reason i am talking to you guys now is that i am still trying to find a way to avoid the fork
[23:49] <thomasvs> ChrisHJW: I was under the impression that you guys were working on a port and that we were to expect patches sometime soon
[23:49] <BBB> thomasvs: read what I just said to ds-work
[23:49] <thomasvs> ChrisHJW: a fork would just be stupid if all you need is a bunch of additional files for visual c
[23:49] <BBB> (17:35:14) BBB: ds-work: I told them to hold off patches
[23:49] <BBB> (17:35:20) BBB: since they use custom vararg macros for now
[23:49] <BBB> (17:35:26) BBB: I told them that'll never go in
[23:49] <BBB> (17:35:43) BBB: read: #define GST_DEBUG_1(x) ..
[23:49] <BBB> (17:35:50) BBB: #define GST_DEBUG_2(x,y) ...
[23:49] <thomasvs> BBB: what did you just say ?
[23:49] <ChrisHJW> thomasvs : our motivation to try to apply these patches was falling to zero when BBB made clear they could never go into main CVS
[23:50] <BBB> I gave alternatives that would go in
[23:50] <BBB> I think it's quite clear that stuff as above is evil
[23:50] <BBB> if that's not clear... well, then I seriously need to change my opinion on win32 programmers ;)
[23:50] <thomasvs> BBB: if it's ifdef'd per platform, why do we care ?
[23:51] <ds-work> It would take about 30 minutes to write inline functions
[23:51] <BBB> thomasvs: because it touches every plugin and makes it use GST_DEBUG_1("debug message"); and GST_DEBUG_4("debug message %s%p%d", bla, foo, bar);
[23:51] <BBB> do you see that this is evil?
[23:51] <BBB> do you *agree* that this is evil?
[23:51] <thomasvs> ChrisHJW: and while I respect BBB's opinion, what sort of person accepts the word of one developer face value ?
[23:51] <thomasvs> BBB: sure I do
[23:51] <thomasvs> BBB: but I can only comment when I know about the specifics :)
[23:52] <thomasvs> ChrisHJW: I'm getting the feeling you guys aren't used to working with open source projects :)
[23:52] <BBB> well, I don't want this sort of stuff inside our plugins, even if it's #ifdef'ed
[23:52] <BBB> I'd rather start using static inline funcs then
[23:52] <BBB> I don't have much against those anyway
[23:52] <thomasvs> ChrisHJW: in this case, it's not that hard to discuss this stuff on the mailing list of the project you're using.
[23:52] <BBB> apart from them taking up lots of space in header files
[23:53] <Tack> BBB: Are there many vararg macros in gstreamer?
[23:53] <thomasvs> BBB: no, sure, I agree with you.  I'm just saying, if this is being discussed between two people instead of between two groups of people, it's a very unconstructive way of working out how to work together
[23:53] <thomasvs> and, I CANNOT believe this problem isn't already solved by other projects.
[23:53] <thomasvs> I mean, g_strdup_printf () works on windows no ?
[23:53] <BBB> tack: yes
[23:53] <BBB> thomasvs: that's a function ;)
[23:53] <ds-work> variadic macros has always been a problem for gstreamer's portability
[23:53] <BBB> I told them to use fuctions in the windows case
[23:54] <Tack> thomasvs: Well, as I said, glib uses #ifdefs and uses vararg macros for compilers that support it, and static functions for those that don't.
[23:54] <ds-work> that's why I removed all of them (except debugging) for 0.8
[23:54] <taaz> does glib have any support for this?
[23:54] <thomasvs> BBB: right, so in windows it can be made a function, no ?
[23:54] <BBB> yes
[23:54] <thomasvs> BBB: ie, with an ifdef in the core source, without having to touch the plugins ?
[23:54] <BBB> yup
[23:54] <BBB> I proposed that several times
[23:54] <thomasvs> BBB: ok, so what's wrong with that for them ?
[23:54] <BBB> somehow, it didn't get through :-p
[23:55] <thomasvs> ChrisHJW: you see why it's stupid to have a single line of communication ???
[23:55] <thomasvs> ds-work: can we make gstreamer.net point to gstreamer.freedesktop.org ?
[23:55] <Tack> BBB: Did they not understand, or disagree?
[23:55] <BBB> tack: I think they mostly ignored it every time I brought it up...
[23:55] <thomasvs> ChrisHJW: where's the gmane link to this discussion ?
[23:56] <BBB> I would have to read through the matroska-devel archives to be sure
[23:56] <ChrisHJW> thomasvs : it was me sending at least 3 emails about this subject to your list ( [email protected] ), with only BBB replying ;) ...
[23:56] <ChrisHJW> thomasvs : the link is given in my last email to gstreamer-devel ;)
[23:57] <BBB> I think they really just want us to stop using vararg macros alltogether and use static inline _only_
[23:57] <BBB> I'm not sure if that's the right solution
[23:57] <Tack> BBB: Profilng might suggest that it is.
[23:57] <Tack> Would be an interesting benchmark.
[23:58] <BBB> I think that for at least gcc, static inline and macros give the same result
[23:58] <Tack> Or pretty damn close, at least.
[23:58] <BBB> at least those sort of claims are being made on linux-kernel
[23:59] <Company> sure
[23:59] <Company> static inline and macros should end up the same
[23:59] <ds-work> thomasvs: yes.  we need to move ownership and the DNS servers first
[23:59] <Company> at least if the compiler does inline
[23:59] <BBB> ds-work: oh, and spyder482 is John Cannon, iirc
[00:00] --- Thu Apr 15 2004
[00:00] <Tack> What happens if you have a recursive function with an inline declaration?  Does the compiler ignore inline?
[00:01] <Tack> Well it must, since I just tried it and it worked and the compiler didn't complain.
[00:01] robUx4 ([email protected]) joined #gstreamer.
[00:01] <robUx4> lo
[00:01] <Company> inline does not force the compiler to inline
[00:01] <Company> inline is just a suggestion
[00:01] <ds-work> Tack: it either inlines it recursively or creates a static function
[00:01] <ChrisHJW> nite guys
[00:01] <robUx4> but in optmisation mode it will do what is best
[00:01] <robUx4> nite ChrisHJW
[00:02] <ChrisHJW> bed time for me ..... robUx4 is in vacation in south america right now, he can stay here for some more time i guess ;)
[00:02] <BBB> sure ;)
[00:02] <BBB> nite :)
[00:02] <Tack> ds-work: Sounds like a pretty smart compiler optimization if it can figure out how many times to inline it when it's being called recursively.
[00:02] <robUx4> yeah, I work here during my vacations ;)
[00:02] <thomasvs> Hi steve
[00:02] LeRoutier ([email protected]) joined #gstreamer.
[00:02] <robUx4> hi
[00:02] <LeRoutier> hello
[00:02] <thomasvs> robUx4: what exactly would you not like about using inline functions for the debugging stuff
[00:02] <thomasvs> ?
[00:02] <robUx4> actually I like inlines :)
[00:04] <ds-work>    This is undocumented and will go when 0.8 comes out. */
[00:04] <Zeenix> robUx4: me too, I need to consult other's code to know how to write macros when i need to ;)
[00:04] <ds-work>  /* This is a workaround so the old debugging stuff of GStreamer 0.6 works.
[00:04] <ds-work>    This is undocumented and will go when 0.8 comes out. */
[00:05] <thomasvs> robUx4: ok, so why fork ? :)
[00:05] <Company> ds-work: remove it, they have been warned :)
[00:05] <ChrisHJW> fine, so we dont need to fork :) ?
[00:05] <robUx4> because we don't want to use GCC under Windows when there are so much better tools
[00:05] <thomasvs> robUx4: sure, but the code is the same, right ? so why do you need a fork ?
[00:05] <thomasvs> robUx4: I thought all you need for visual c is a bunch of project files ?
[00:06] <robUx4> and you refused the patches we made to make GStreamer work with these tools
[00:06] <thomasvs> we refused ?
[00:06] <thomasvs> I didn't see us refuse those
[00:06] <robUx4> no, MSVC doesn't support : #define MACRO(x,y,...)
[00:06] <robUx4> BBB: probably did
[00:06] <thomasvs> robUx4: we just said we are ok with using inlined functions for that on windows :)
[00:06] <thomasvs> robUx4: there is more than one developer working on gstreamer
[00:07] <robUx4> I know
[00:07] <thomasvs> robUx4: and, what he said is correct - a patch that causes us to have to change ALL code to replace GST_DEBUG with GST_DEBUG_x where x is the number of arguments is not the way to fix this
[00:07] <thomasvs> the way to fix is to use inline functions on windows for GST_DEBUG
[00:07] markey ([email protected]) left irc: "nite"
[00:07] <robUx4> yeah, I didn't make that change. too bad Jory is not here
[00:08] <thomasvs> robUx4: it's not uncommon for an open source project to refuse a patch based on the fact that it's not the right way to fix it you know :)
[00:08] <robUx4> if BBB's workaround is OK, then it's fine
[00:08] <thomasvs> robUx4: all that means is that it needs some more discussion.
[00:08] <thomasvs> robUx4: not accepted != rejected
[00:08] <ChrisHJW> over time, maybe people wont include the vararg macros anymore to make the plugins compile on both versions, we'll see ;)
[00:08] <robUx4> but in the end I don't understand why using vararg macros, but... ;)
[00:09] <BBB> because we like them ;)
[00:09] <thomasvs> robUx4: it works for the vast majority of platforms we currently aim at working on
[00:09] <BBB> ChrisHJW: the plugins will work with both, the plugins don't need to be touched
[00:09] <thomasvs> robUx4: nobody gains by having you guys work on your own.  it's much preferable if you guys become part of the team.
[00:09] <Company> varargs macros are faster when the compiler doesn't understand inline
[00:09] <robUx4> thomasvs: I know but from what I understand from Jory noone was willing to make a decent proposition
[00:09] <thomasvs> robUx4: afaict nobody knew that you guys considered it a rejection :)
[00:10] <robUx4> thomasvs: I totally agree. otherwise I wouldn't tell you and we would make our fork in our corner
[00:10] <ds-work> variadic macros are part of the language, so unless there's a good reason, we'll use them
[00:10] Action: ChrisHJW is glad people start talking to each other again and goes finally to bed now
[00:10] <ChrisHJW> nitie nite
[00:10] <thomasvs> robUx4: ok, so let's do that better then.
[00:10] <robUx4> again, Jory is the guy to ask. I'm not home and don't have a Windows machine here
[00:10] <robUx4> (at least not mine)
[00:11] <thomasvs> robUx4: well, can you get him to mail his concerns to our mailing list then ?
[00:11] <thomasvs> ds-work: is there a way to detect if a compiler supports variadic ?
[00:11] <BBB> AC_TRY_COMPILE()?
[00:11] <robUx4> he surely will
[00:11] <Company> thomasvs: there are glib macros defined
[00:11] <ds-work> thomasvs: we already do
[00:11] <ds-work> thomasvs: using glib
[00:12] <thomasvs> ds-work: ok, so right now we just define GST_DEBUG to null on those ?
[00:12] <BBB> oh,right, G_HAVE_GNUC_VARARGS
[00:12] Nick change: mxpxpod -> mx|gone
[00:12] <ds-work> thomasvs: yes
[00:12] <thomasvs> robUx4: where are you in argentina ?
[00:12] <ds-work> thomasvs: er, maybe.  it might just fail
[00:12] <thomasvs> that needed a comma, but hey
[00:13] <Tack> This is interesting.  A very crude benchmark mind you, but testing a macro versus static inline that basically just increments a value, a static inline is 6 times faster than a macro with -O2.
[00:13] <Tack> Quite counter-intuitive.
[00:14] <Zeenix> is Uraeus on his world-tour?
[00:14] <ds-work> not until June
[00:14] <thomasvs> Zeenix: not yet
[00:14] Action: BBB gone
[00:14] <BBB> bbl
[00:14] #gstreamer: mode change '+o thomasvs' by ChanServ!ChanServ@services.
[00:15] BBB ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[00:15] <Tack> How is this possible.  Makes no sense. :)
[00:15] <robUx4> thomasvs: in Colombia, Bogota
[00:15] Action: Zeenix does'nt understand why Uraeus is so affraid to come to pk
[00:15] Topic changed on #gstreamer by [email protected]: Test prereleases - http://gstreamer.freedesktop.org/src/gstreamer/pre/gstreamer-0.8.0.2.tar.gz and http://gstreamer.freedesktop.org/src/gst-plugins/pre/gst-plugins-0.8.0.2.tar.gzif you're trying to debug gst-launch and fail, try gdb gst-launch-0.8 | Media playback tracker bug: #138435 | more april foolery: #138831 (a palindrome, even) | someone release gst-monkeysaudio, there's no release compatible with 0.8.0
[00:16] <ds-work> Tack: look at the generated assembly.  btw, ignore the results of your test, since there are lots of factors that could contribute
[00:16] <ds-work> thomasvs: lots of the topic could be trimmed
[00:17] <thomasvs> ds-work: sure, tell me what I can throw out
[00:17] <ds-work> s/more april .*//
[00:17] Topic changed on #gstreamer by [email protected]: Test prereleases - http://gstreamer.freedesktop.org/src/gstreamer/pre/gstreamer-0.8.0.2.tar.gz and http://gstreamer.freedesktop.org/src/gst-plugins/pre/gst-plugins-0.8.0.2.tar.gzif you're trying to debug gst-launch and fail, try gdb gst-launch-0.8 | Media playback tracker bug: #138435
[00:17] <ds-work> s/gzif/gz | if/
[00:18] Topic changed on #gstreamer by [email protected]: Test prereleases - http://gstreamer.freedesktop.org/src/gstreamer/pre/gstreamer-0.8.0.2.tar.gz and http://gstreamer.freedesktop.org/src/gst-plugins/pre/gst-plugins-0.8.0.2.tar.gz | if you're trying to debug gst-launch and fail, try gdb gst-launch-0.8 | Media playback tracker bug: #138435
[00:18] Action: thomasvs puts the apple within reach
[00:18] #gstreamer: mode change '+o ds' by [email protected]
[00:18] #gstreamer: mode change '-o thomasvs' by [email protected]
[00:19] #gstreamer: mode change '+o ds-work' by ChanServ!ChanServ@services.
[00:19] #gstreamer: mode change '-o ds' by [email protected]
[00:19] #gstreamer: mode change '+o ds' by ChanServ!ChanServ@services.
[00:19] #gstreamer: mode change '-o ds-work' by [email protected]
[00:19] <ds-work> er, nevermind
[00:20] <thomasvs> what a wasteful display of power
[00:21] Action: ds-work kicks thomasvs
[00:22] <robUx4> :)
[00:24] <thomasvs> ds-work: face it, you're growing old.  you're not a real hax0r anymore
[00:24] <Tack> ds-work: True, it did some funky optimizations. :)
[00:26] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[00:27] <Tack> Compiler seems to ignore inline, though, unless at least -O1 is called.
[00:27] <ds-work> yes
[00:27] Action: ds-work has a vast store of underutilized optimization information
[00:28] <Tack> With -O1, the assembly looks pretty much the same.
[00:28] trow ([email protected]) left irc: "Leaving"
[00:28] <kmaraas> thomasvs, I think gnome-volume-control+alsa is a bit broken
[00:28] <Tack> And indeed, so are the results.  -O2 produced some interesting optimizations.  With the inline function version, instead of iterating N times and adding 1 to the counter, it iterated N/10 times and added 10.
[00:29] <Tack> With the macro version it didn't see that optimization.
[00:29] <ds-work> probably the same with explicit code
[00:30] <ds-work> gcc sometimes assumes that you actually indend to have delay loops if you write them
[00:30] ChrisHJW ([email protected]) left irc: Success
[00:31] <Tack> Really?  I'd not have thought. :)
[00:31] <jdahlin> __attribute__(const) is also cool
[00:32] <ds-work> so is __attribute__(pure)
[00:32] <thomasvs> kmaraas: ?
[00:32] <Tack> Well, explicit code should be identical to the macro, since the preprocessor is done before assembling.
[00:39] jdahlin ([email protected]) left irc: "ZZZZ"
[00:48] yippi ([email protected]) joined #gstreamer.
[00:52] kmaraas ([email protected]) left irc: Client Quit
[00:58] <ds-work> ok, we have inlines for debugging functions now
[01:00] <Company> __attribute__(const) is not cool
[01:00] <Company> because -O2 optimizes calls to such functions away
[01:00] <LeRoutier> ouch, got gst-player segfaulting when playing a .dv file
[01:00] <Company> which causes gst_caps_get_type to not be called in gst_init
[01:00] spyder482 ([email protected]) joined #gstreamer.
[01:01] <Company> (which i wanted to file as a bug or fix but haven't iirc)
[01:10] benow ([email protected]) left irc: "Guiding the meefalo to glorious meef free living"
[01:14] sublett ([email protected]) left irc: "I quit, therefore I was"
[01:17] yippi ([email protected]) left irc: "Client exiting"
[01:18] robUx4 ([email protected]) left #gstreamer ("I like core dumps").
[01:18] robUx4 ([email protected]) joined #gstreamer.
[01:19] <robUx4> oops
[01:31] jcsston ([email protected]) joined #gstreamer.
[01:31] <robUx4> here is Jory :)
[01:31] <jcsston> hi
[01:35] <robUx4> thomasvs: are you around ?
[01:38] <LeRoutier> anyone know where the error is in this pipeline ? 
[01:38] <LeRoutier> gst-launch --gst-scheduler=basicgthread dv1394src ! dvdec ! ffmpegcolorspace ! ximagesink
[01:49] benow ([email protected]) joined #gstreamer.
[01:50] <Zeenix> hi benow
[01:53] sub_pop ([email protected]) left irc: "Client exiting"
[01:55] sub_pop ([email protected]) joined #gstreamer.
[01:57] Rotty ([email protected]) left irc: "Bye"
[02:00] <benow> heya Zeenix
[02:03] <Zeenix> benow: which distro you use?
[02:06] <LeRoutier> good night ppl
[02:06] LeRoutier ([email protected]) left irc: "Leaving"
[02:09] BBB ([email protected]) joined #gstreamer.
[02:10] <BBB> dus... can I still commit?
[02:10] <BBB> I have MS RLE working locally
[02:11] pb_ (~pb@2002:5160:4550:1:240:95ff:fe30:aa84) left irc: "Client exiting"
[02:14] iain ([email protected]) joined #gstreamer.
[02:14] spyder482 ([email protected]) left irc: "Leaving"
[02:23] Action: Company kicks libc
[02:23] <Company> strchr should return a const
[02:26] <Zeenix> Company: libc can't feel it, kick RMS :)
[02:28] Action: BBB hates the CVS freeze
[02:28] Action: BBB wants to commit fixes
[02:28] <taaz> are they fixes that should be in 0.8.1?
[02:29] Action: BBB thinks so
[02:29] <BBB> trying to think if thomasvs would approve
[02:29] <BBB> he probably wouldn't know and ask me if they could possibly break something
[02:29] <BBB> and I'd lie and say no
[02:29] <BBB> and I'd commit
[02:29] <BBB> and the whole world would be happy
[02:29] <Company> are they going to make things worse?
[02:29] <BBB> nah, they fix MS RLE playback
[02:30] <BBB> thanks to the sample file provided by leroutier
[02:30] <Company> are you going to make thomasvs happy again if they break something?
[02:30] <BBB> I'll revert them if they break something
[02:30] <Company> we'll probably find out after 0.8.1 is out ;)
[02:30] Action: taaz notes this is what a test suite is for ;)
[02:31] <BBB> this is new :p
[02:31] <BBB> you could call it a feature or a bug, depending on your pov
[02:31] <Company> i'd say go ahead if you think it should be in, but i'm not the release manager
[02:32] <BBB> I'll think of a good excuse ;)
[02:32] Action: BBB first tests current CVS
[02:32] <taaz> um.. so i think i have the debs for 0.8.0 done now. ;)
[02:32] <BBB> did you guys test the prereleases?
[02:32] <BBB> i.e. should I?
[02:32] <BBB> taaz: upload
[02:32] <BBB> and update to 0.8.1 within a week
[02:33] <taaz> BBB: it may take a week for them to pass the gatekeepers of the NEW queue
[02:33] <Company> BBB: i'm always running cvs HEAD, so yes, I guess i'm testing prerels
[02:33] <BBB> Company: I meant things like installation and so on
[02:34] <BBB> that's ususally the only thing I test with prereleases
[02:34] <Company> taaz: good thing is that 0.8.0 => 0.8.1 is "only" the kio plugin :)
[02:34] <BBB> "does make install work?"
[02:34] <Company> BBB: i run installed ;)
[02:34] <BBB> the rest is element fixes, indeed
[02:34] <BBB> ah
[02:34] <taaz> Company: what is kio anyway?
[02:34] <BBB> taaz: do you package gst-ffmpeg?
[02:34] <Company> taaz: KDE's gnome-vfs
[02:34] <BBB> KDE's gnomevfs
[02:34] Action: BBB slaps company
[02:34] <BBB> ;)
[02:34] <Company> taaz: it doesn't work very well yetr though
[02:34] <taaz> BBB: haven't gotten to that yet.  shouldn't be hard though.
[02:35] <BBB> it's only one file...
[02:35] <taaz> Company: what are the lib deps?  am i going to need to make a -kio package?
[02:36] <Company> taaz: kdelibs
[02:39] <taaz> oh crap, i lied.  my final build decided to not actually include the plugin libs.  whoops
[02:39] Action: taaz wonders how that happened...
[02:40] Action: BBB suggests it's all taaz' fault
[02:40] <BBB> ;)
[02:40] <BBB> j/k :p
[02:40] <taaz> it worked an hour ago just fine... dunno what i changed ;)
[02:41] <taaz> and i just deleted the build log.. grr
[02:46] <benow> Zeenix: gentoo
[02:52] 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 


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