From: IRC <wt...@us...> - 2003-12-02 11:23:10
|
******************************************************************* [03:01] sxpert (~sx...@sx...) left irc: Read error: 110 (Connection timed out) [03:26] harshy (~ha...@dh...) joined #gstreamer. [03:53] Zeenix (~zak@203.135.11.21) left irc: "The whole problem with the world is that fools and fanatics are always so certain of themselves, but wiser people are so full" [03:53] Marsupilami23 (~Mar...@15...) joined #gstreamer. [20:34] _gst_newt_ joined #gstreamer. [20:54] BBB (~rb...@st...) joined #gstreamer. [20:57] Action: BBB is really annoyed by all this video issues [20:57] <BBB> someone broke the ffmpeg mpeg4 decoder, apparently [20:57] <BBB> it won't decode anything anymore [20:57] <BBB> :/ [20:59] Company (~Co...@p5...) joined #gstreamer. [21:04] <yves_> hi BBB [21:04] Nick change: yves_ -> Makki [21:04] <BBB> hi [21:04] <BBB> hey company [21:05] Action: Company is not there [21:05] <Company> at least in theory [21:06] <BBB> right [21:06] <Makki> mind needs fixing ? [21:06] <BBB> Company: any documentation on the new metadata system? [21:06] <Company> BBB: the code [21:06] <Company> shit, i still have to write it... [21:06] <BBB> eek [21:07] Action: BBB kicks company [21:07] <Company> the stuff in the mad dir has it [21:07] Action: Makki is away: done working [21:07] <Company> and plugins/gst/tags/ [21:07] <BBB> metadata is the only thing left in the matroska demuxer [21:07] <BBB> and rifflib/avidemux is finished too [21:07] <BBB> but needs metadata [21:07] <BBB> ;)_ [21:08] <Company> and you can ask teuf [21:08] <Company> ;) [21:09] <Company> cairo is really cool btw [21:09] <Company> only problem is it's sloooooow [21:09] <Company> and doesn't do YUV yet [21:10] <Company> though that shouldn't be too hard (but even more slow ;) [21:10] <BBB> cairo? [21:10] <Company> www.cairographics.org [21:11] <Company> image compositing library abstracting the backend [21:11] <Company> does images, X, ps, ... [21:11] <Company> and i wanted to do textoverlay and TV stuff with it [21:23] Nick change: TD -> TD_ [21:24] <dolphy> BBB: if i set format props on a rgb caps [21:24] <dolphy> BBB: will it come back on sink connect ? [21:24] <BBB> probably not [21:25] <dolphy> BBB: when i get xv supported formats i get a format id [21:25] <dolphy> BBB: i need that format id [21:25] <dolphy> BBB: so for yuv it is the format props directly [21:25] <dolphy> BBB: so i get it back [21:25] <dolphy> BBB: but for RGB no such thing [21:25] <BBB> you need to retrieve it from the caps, sorry [21:25] <BBB> that's how other plugins such as v4l, v4l2, ffmpeg etc. do it [21:25] <dolphy> BBB: how ? [21:26] <BBB> simply go through the list of all known caps [21:26] <BBB> with the format id [21:26] <BBB> and if you've found a caps that is similar to your [21:26] <BBB> use that format id [21:26] <dolphy> so i have to store every format and it's caps [21:27] <BBB> yes [21:27] <dolphy> and then on sinkconnect search through my list [21:27] <BBB> idd [21:27] <dolphy> to get the similar caps [21:27] <dolphy> and get the format id [21:27] <dolphy> that sux a little [21:27] <dolphy> why caps intersection is not letting optional caps in the intersected one ? [21:28] giard (~jon...@19...) joined #gstreamer. [21:30] <dolphy> BBB: i have an idea :) [21:30] <dolphy> BBB: am i getting back the caps name in sinkconnect ? [21:31] <BBB> no [21:31] <BBB> you cannot do hacks, you really have to do a list [21:31] <BBB> sorry [21:32] <dolphy> :-( [21:32] <dolphy> that sux [21:32] <BBB> sorry... but it's really not that slow [21:33] <BBB> it's just 10-20 lines of code [21:33] <dolphy> that doesn't sound logical [21:33] <dolphy> having to intersect every supported caps [21:33] <dolphy> with the one sinkconnect gives me just to gather a small integer [21:33] <BBB> your computer can do 1-2 billion calculations per second [21:34] <BBB> sorry, but this is just the way it si :/ [21:34] <dolphy> which other plugin than xvideosink is doing that [21:34] <dolphy> ? [21:35] <BBB> a conversion from gstreamer system to internal system/. [21:35] <BBB> ? [21:35] <BBB> ehm [21:35] <BBB> all of them ;) [21:35] <BBB> ffmpeg uses gstffmpegcodecmap.c for that [21:35] <BBB> v4l/v4l2 have functions in their gstv4l{,2}src.c file [21:40] <dolphy> BBB: gst_caps_append is not stealing the ref of the second caps right ? [21:41] <BBB> 'stealing'? [21:41] <BBB> yeah, it does [21:41] <dolphy> BBB: in gst_caps_append (caps1, caps2); you have to unref caps2 after ? [21:41] <BBB> no [21:41] <dolphy> BBB: it s merging then ? [21:41] deadChip (~Fi...@di...) joined #gstreamer. [21:41] <BBB> it basically just does gst_caps_last (caps1)->next = caps2; [21:41] <dolphy> i see [21:41] <dolphy> ok so i will copy each caps before appending to a list [21:43] <dolphy> BBB: when gst_caps_intersect returns != NULL with the a caps and the one from sinkconnect i m 100% sure that's the same one ? [21:43] <BBB> uhm, yes [21:43] <BBB> but intersect is expensive [21:43] <BBB> doing gst_caps_get() on the relevant properties and comparing those manually is probably faster [21:44] sisob (~sisob@213.94.251.99) joined #gstreamer. [21:44] <dolphy> BBB: boah i have 1 or 2 billion spare operations ;-p [21:44] <sisob> is there any trick to getting gst-register-0.7 working ? [21:46] <BBB> dolphy: :D [21:46] <BBB> sisob: ehm, don't know... what's wrong + what's the backtrace? [21:46] <sisob> [sisob@sisob gstreamer]$ GST_PLUGIN_PATH=/home/sisob/Software/Gnome26/lib/gstreamer-0.7 sudo gst-register-0.7 [21:46] <sisob> rebuilding global_registry (/home/sisob/Software/Gnome26/var/cache/gstreamer-0.7/registry.xml) [21:46] <sisob> rebuilding user_registry (/root/.gstreamer-0.7/registry.xml) [21:46] <sisob> loaded 0 plugins with 0 features [21:46] Zeenix (~zak@203.175.73.170) joined #gstreamer. [21:46] Company (~Co...@p5...) left irc: Read error: 110 (Connection timed out) [21:47] <sisob> :( [21:47] <dolphy> BBB: huhu you mean making a get on every props [21:47] <dolphy> BBB: and comparing them manually [21:47] <sisob> BBB: there are loads of plugins installed [21:47] <Zeenix> hello [21:48] <BBB> dolphy: yeah, or use gst_caps_get () [21:48] <BBB> that's just one call [21:48] <BBB> isn't that --gst-plugin-path=/path/..? [21:50] <sisob> BBB: i think that both should work, neither do [21:50] <BBB> I don't know [21:51] <BBB> I always use the second [21:52] <sisob> :( [21:52] <sisob> BBB: what prefix you using? [21:53] <BBB> GST_PLUGIN_PATH=/home/sisob/Software/Gnome26/lib/gstreamer-0.7 sudo gst-register-0.7 - I think that env. vars are cleared before sudo gets root [21:53] <BBB> sisob: default, so /usr/local [21:54] <giard> register doesn't have to be run as root does it? [21:54] <dolphy> BBB: i will do gst_caps_intersect only if _caps_get_mime returns rgb [21:54] <dolphy> BBB: this way yuv goes straight [21:54] <dolphy> BBB: and i m just checking RGB caps [21:54] <dolphy> BBB: do you think that's ok ? [21:54] <sisob> BBB: the sudo thing wan't it, i have mine installed into a custom prefix, could be it [21:55] <BBB> dolphy: perfect :) [21:55] <BBB> giard: no [21:56] <dolphy> BBB: could you tell me why rgb does not have fourcc format ? [21:56] <dolphy> BBB: in the caps ? [21:57] <BBB> because we don't need it [21:57] <BBB> caps is supposed to be simple [21:59] <dolphy> BBB: and framerate is ? [21:59] <BBB> framerate is a stupid mistake of mine that should have been optional instead of mandatory [22:00] <BBB> if you don't look at it and remove it from a sink template caps, it'll still work, so go ahead and do that if you want [22:00] <dolphy> well i don't mind having framerate :) [22:00] <dolphy> but wait [22:00] <BBB> you can't use it, can you? [22:00] <BBB> it's for headers etc. [22:00] <BBB> such as AVI, matroska, ... [22:00] <dolphy> i thing XV formats are just fourcc [22:01] <dolphy> so if your xv adaptor supports RGB [22:01] <dolphy> the format id it returns is probably GST_MAKE_FOURCC ('R','G','B',''); [22:01] <dolphy> right ? [22:01] <BBB> no [22:01] <dolphy> nothing different [22:01] <BBB> for RGB, it will be RV16 on my video card ;) [22:02] <BBB> you really want to make a function that does gst_get_xv_fourcc_from_caps (); [22:02] <BBB> as I said, ffmpeg, v4l, v4l2 et all have that too [22:02] <BBB> it's normal to have that [22:02] <BBB> even sdl has that [22:02] <dolphy> snif [22:02] <dolphy> :'-( [22:02] Action: dolphy does that [22:02] <BBB> sorry, the world is hard and hates us programmers [22:02] <BBB> ;) [22:07] <dolphy> BBB: gst_caps_intersect will unref my caps ? [22:07] <dolphy> BBB: i mean if i walk through my caps list and intersect [22:07] <dolphy> BBB: will i be able to do it a second time with the same caps list ? [22:07] <BBB> I don't know [22:07] <BBB> check the source :) [22:08] Action: BBB checs [22:08] <dolphy> i m still a bit lost with those caps_sink concept [22:08] <BBB> they're not unreffed [22:08] <BBB> me too [22:08] <BBB> it hasn't become easier [22:08] <BBB> but anyway [22:08] <BBB> so yes, you can intersect twice [22:09] <dolphy> good [22:14] TD_ (~mi...@bd...) left irc: "leaving" [22:14] giard (~jon...@19...) left irc: Read error: 104 (Connection reset by peer) [22:15] sisob (~sisob@213.94.251.99) left irc: "I like core dumps" [22:16] walters (wa...@ve...) joined #gstreamer. [22:22] sublett (~rv...@21...) joined #gstreamer. [22:22] <Zeenix> BBB: you are undergrad. student? [22:25] <BBB> I'm studying for my MSc [22:25] <BBB> is that undergrad? [22:28] <Zeenix> BBB: no its not. have you ever prepared a 'research proposal' ? which is required for M.Phil/Ph.D [22:28] Company (~Co...@p5...) joined #gstreamer. [22:29] <teuf> yay Company :) [22:30] <BBB> Zeenix: I've done that a year ago when I started my graduation project [22:30] <BBB> I've almost finished the project now [22:30] <Company> yay me [22:30] <teuf> Company: can you have a quick look at my flac metadata patch so that I know if I forgot stuff or made something wrong ? [22:30] <BBB> just need one or two small results and then I'm done [22:30] <BBB> working on the thesis now [22:31] <Company> teuf: just commit it - if something breaks, someone will notice [22:31] <BBB> I'll use that for avi, too [22:31] <Company> (or not - flac ddidn't work for how long? ;) [22:31] <BBB> ain: will you rewrite wavparse to use rifflib? [22:31] <Company> teuf: i'm with friends this weekend [22:31] <teuf> Company: I already committed it [22:31] <Company> teuf: i can look it over sometime next week maybe [22:32] <teuf> ok cool :) [22:32] <Zeenix> BBB: can you guide me how to do that. my problem is that i am tooo objective-minded. [22:32] Action: Company declares teuf the tagging guru [22:33] <teuf> Company: there are still some parts I don't understand :) [22:33] <BBB> Zeenix: wavparse rewriting? [22:33] <BBB> Zeenix: if you want to, sure [22:34] <BBB> Zeenix: wait a few days for me to commit the new avidemux and rifflib [22:35] <BBB> Zeenix: or do you mean how to do a project? ;) [22:35] <BBB> that's simple [22:35] <BBB> just do what you studied to do in the years before that [22:35] <Zeenix> BBB: nooooo , i mean writing the research proposal [22:36] <BBB> ah ;) [22:36] <teuf> Company: what does gstid3tag.c does compared to the metadata handling from gstmad.c and gstid3types ? [22:36] <BBB> for me, it was simple: write down what you want to do, what you hope to find, theoretical background and theory on why you expect that, plus methods description [22:37] leif (~le...@ad...) joined #gstreamer. [22:37] <Zeenix> BBB: are you good at subjective stuff? [22:38] <Company> teuf: id3tag is a metadata extractor/setter while id3types is an id3 "demuxer" and mad is an mp3 decoder [22:38] <BBB> depends on the subject... [22:38] <BBB> I've written quite a few theses for philosophy, most of that was very subjective [22:38] <BBB> some papers were good [22:38] <BBB> but some were really bad [22:38] <BBB> I was especially bad at critisizing Nietzsche [22:39] <BBB> what's id3tag? [22:39] leif__ (~le...@ad...) left irc: Read error: 60 (Operation timed out) [22:39] <Company> teuf: id3tag is a metadata extractor/setter while id3types is an id3 "demuxer" and mad is an mp3 decoder] [22:40] <Company> s/teuf/BBB/ :) [22:40] <Zeenix> BBB: what is Nietzsche? [22:40] <BBB> a philosopher [22:40] <Company> teuf: theoretically id3tag and id3types could be the same code. i'm trying to do that right now [22:41] <BBB> Company: those are all the same [22:41] <teuf> Company: hmm, and how exactly is a metadata extractor different than using the mad decoder (I have the feeling I already asked, but I forgot :-/) [22:41] <teuf> ? [22:41] <BBB> why separate elements for all those? [22:41] <BBB> I don't get it [22:41] <BBB> we need to remove the id3 code from mad [22:41] <BBB> how are id3tag and id3types different? [22:42] <BBB> anyway, I'm gone [22:42] <BBB> bbl [22:42] <BBB> probably tomorrow [22:42] <BBB> bye [22:42] BBB (~rb...@st...) left irc: "Client exiting" [22:42] <Company> the problem with id3types is that it supports "any" output [22:42] <Company> and i haven't found a way to support "any" and something else :) [22:44] <Company> though that should be possible [22:45] RedFox (~redfox@ABayonne-105-2-1-201.w81-50.abo.wanadoo.fr) joined #gstreamer. [22:45] <RedFox> Hi [22:45] <Company> hi [22:45] <teuf> hey RedFox :) [22:45] <RedFox> It seems that gstreamer 0.7.2 fails to compile on PowerPC, at least for me :( [22:46] <RedFox> ppc-redfox-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I. -I../../.. -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -I/usr/include/libxml2 -pthread -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I../../.. -Wall -DGST_DISABLE_DEPRECATED -Wa,-mregnames -g -I../../../libs -I../../../include -Wall -pipe -O3 -fsigned-char -fomit-frame-pointer -ffast-math -mcpu=750 -DNDEBUG -DG_DISABLE_ASSERT -MT libgstcontrol_0.7_la-dplinearinterp.lo -M [22:46] <RedFox> D -MP -MF .deps/libgstcontrol_0.7_la-dplinearinterp.Tpo -c dplinearinterp.c -o libgstcontrol_0.7_la-dplinearinterp.o >/dev/null 2>&1 [22:46] <RedFox> make[4]: *** [libgstcontrol_0.7_la-dplinearinterp.lo] Error 1 [22:47] <teuf> RedFox: nothing more ? [22:47] <RedFox> Ouupsss, I think I found the bug !! [22:47] <RedFox> I have no space on my hard disk :((( [22:47] <teuf> hehe ;) [22:48] <dolphy> BBB: ready for a test ? [22:48] <RedFox> Sorry , It's late for me :/ [22:48] <teuf> nice to see you there [22:48] <dolphy> raah he left [22:48] <dolphy> Company: ping? [22:48] <teuf> RedFox: did you hear about this gnome-fr list at gnome.org ? [22:48] <Company> dolphy: pong [22:48] <RedFox> teuf: yes [22:49] <RedFox> I have strider on a private channel [22:49] <teuf> ok cool [22:49] <teuf> great [22:49] <dolphy> Company: xvinfo please [22:50] <Company> huh? [22:50] <dolphy> Company: the output of the xvinfo command on your laptop [22:50] <dolphy> Company: for the BadMatch problem [22:51] <Company> dolphy: i think that was my home box [22:51] <dolphy> Company: BBB has the same problem [22:51] <dolphy> Company: it's because your XV Adaptor supports RGB image formats [22:51] <dolphy> Company: and xvimagesink was not supporting RGB correctly yet [22:51] <dolphy> Company: i just commited a fix for that [22:51] <dolphy> Company: but i can't test it [22:51] <dolphy> Company: as my ATI only has YUV [22:52] <Company> both my cards are ATI and don't support RGB [22:53] Action: teuf tries to port sound-juicer to the new metadata api :) [22:53] <Company> teuf: great :) [22:54] <dolphy> Company: ah weird then [22:54] <Company> teuf: you could try to implement musicbrainz support in cdparanoia - that way you could get rid of all metadata code in SJ [22:55] <teuf> Company: yeah, I thought about that, but I don't really feel like implementing that ;) especially since it may be tricky to deal with multiple matches for the same album [22:55] <Company> true [22:56] <Company> we need a function gst_bin_find_element_by_interface [22:57] <Company> that way you could have SJ use user-defined pipelines and lookup the tagsetter from that [22:57] <teuf> Company: I'm supposed to used gst_tag_setter_add for example to set metadata from soundjuicer ? [22:57] <Company> teuf: yeah [22:58] <teuf> tag being kept when transcoding with gst-launch really rocks :) [22:58] <Company> teuf: that interface hasn't been tested real-life either, so if you miss something, feel free to add it [22:58] <Company> yeah, i liked that, too :) [22:59] <teuf> Company: gst_tag_setter_merge is meant more for elements than for apps ? [23:00] <Company> teuf: no, elements shouldn't add stuff to the tagsetter tag at all [23:00] <Company> teuf: it's meant for apps that adopt the GstTagList structure for keeping metadata [23:01] <Company> teuf: the idea is that apps use the tagsetter interface to attach tags to an element and the elements reads those out whenever it's going to write metadata but doesn't change the tagsetter tags itself [23:23] Nick change: sxpert -> waloo_ [23:23] Nick change: waloo_ -> sxpert [23:26] sxpert (~sx...@sx...) left irc: "Client exiting" [23:28] seb128 (~seb128@ANancy-111-1-7-76.w81-250.abo.wanadoo.fr) joined #gstreamer. [23:28] rgammon (~Ry...@ip...) joined #gstreamer. [23:30] <RedFox> OK, there is no bug in gstreamer 0.7.2. It just need 100MB of space to compile ! [23:31] <walters> is sf cvs borked? [23:35] mathrick (~mathrick@Zietka-18.a-inter.net) joined #gstreamer. [23:36] Miko5881 (~Mik...@68...) joined #gstreamer. [23:39] RedFox (~redfox@ABayonne-105-2-1-201.w81-50.abo.wanadoo.fr) left irc: Read error: 104 (Connection reset by peer) [23:39] <teuf> walters: doesn't seem to work here [23:39] <teuf> grr, I need to recompile a kernel to test sound-juicer :( [23:44] walters (wa...@ve...) left irc: "out" [23:44] <deadChip> question, our app bmp (just like xmms) has prop. plugins, and the plugin interface provides a bundling function input_check_file(), which in turn calls all the input plugins' function is_our_file(), which returns TRUE when the plugin is capable of playing it [23:44] <deadChip> is there something like that possible with gst too? [23:45] <deadChip> querying gst if it can play a certain file? [23:46] <Company> yes and no [23:46] <Company> we are able to detect a type [23:46] <Company> but we're not able to tell if we can play that type [23:47] <deadChip> no way to tell if a decoder element for the file i want to play exists? [23:48] <Company> yes, that's possible [23:48] <Company> in theory [23:49] <Company> because AVIs may contain unknown codecs [23:49] m_wheels (~sc...@ds...) joined #gstreamer. [23:49] <deadChip> hmmm ok [23:49] <deadChip> but good it's generally possible [23:50] <deadChip> i wouldn't like the idea that we might need extra code to do that :| [23:50] <deadChip> err [23:50] <Company> you could just query all decoders if they support a given type and if you find one you could assume it's possible [23:50] <deadChip> i mean to use the libs ourselves doubled [23:50] <deadChip> yeah that's what i thought [23:50] <Company> GStreamer uses autoplugging [23:50] <deadChip> heh i just started to read the devel docs but it seems there can't be anything simpler to code with :D [23:51] Nick change: m_wheels -> wheels [23:51] <Company> so it detects the type automagically and connects the needed decoders [23:52] <deadChip> well we need the first check as well since we have to determine if we should add a file to the playlist, or not [23:53] <Company> hm, yeah [23:53] <Company> that's difficult [23:54] <Company> though for audio only it'll probably be enough to just look for a decoder for $format [23:59] <mathrick> Company: how expensive will it be to construct pipeline for each one? [00:00] --- Sat Nov 29 2003 [00:03] <Company> just for typefinding? [00:03] <Company> or for what? [00:04] <deadChip> thanks Company [00:04] <mathrick> Company: yeah, just for that [00:04] <Company> mathrick: you can reuse pipelines (in that case all bugs for reusing are fixed ;)) [00:05] <Company> mathrick: use gst-typefind-0.7 <file1> <file2> ... to test [00:06] <Company> mathrick: i think gst-typefind-0.7 *.mp3 used 2 secs. per 1000 songs without file aceess last i tried, but test yourself, that number was from memory [00:06] <Company> deadChip: np :) [00:06] <deadChip> : ) [00:07] sxpert (~sx...@sx...) joined #gstreamer. [00:08] <mathrick> Company: I wonder how well could it work for video - it will be something like - first typefind, and if type is known, then try to connect pipeline for it to try if it plays, right? [00:09] <mathrick> Company: but, it would probably work really bad for network based sources :\ [00:09] <deadChip> do you mean streams or things like NFS/SMB? [00:09] <Company> mathrick: i haven't really thought about it, but I guess you'd go multithreaded and have a low priority thread in the background doing that [00:10] <Company> mathrick: you don't need that much data for finding that out - like the first 100kB or so [00:10] <Company> 10kB probably [00:10] <mathrick> deadChip: both, streams could orders of magnitude worse, tho [00:10] <Company> maybe even 1kB is enough, depends on the video format I guess [00:10] <deadChip> hmm yeah didn't think about network yet [00:11] <deadChip> well.. [00:11] <deadChip> a stream has to be autoplugged-only then [00:11] <deadChip> e.g. adding URLs to the playlist without prior checking [00:11] <deadChip> and if it's just remote filesystem, well then it may be slower yes [00:11] <deadChip> systems* [00:12] <mathrick> deadChip: so it won't be autoplugged then [00:12] <Company> yeah, just add all of them and check in the background [00:13] <deadChip> it will work more or less in 2 stages: check the file at playlist add, and then autoplug it when it should be played. that is a bit wacky, since in the 2nd part we simply "trust" the result of the first, but that is almost only a theoretical prob [00:13] <deadChip> yes that's how it works now [00:13] <mathrick> is it possible to check if pipeline's source is network based? [00:14] <Company> not yet [00:15] <Company> reminds me of something... [00:16] <Company> teuf: could you look at the core's gsturi.[ch] and suggest additions that might be useful in gstreamer (like type of source "local", "net", ...) [00:16] <Company> teuf: since i wanted to model it close to gnome-vfs and you know a lot about that [00:16] <teuf> maybe "a lot" is a bit much [00:17] <teuf> differentiating "local" "net" and stuff can be tricky with nfs and things like that [00:17] hadley (~hadley@ACA7D41D.ipt.aol.com) joined #gstreamer. [00:18] <deadChip> how about streams are "net" and nfs/smb and the like "remote"? [00:19] <mathrick> deadChip: I think it will be difficult if you are on nfs - it's mounted like any other fs [00:19] <teuf> deadChip: you are generally interested to know whether it's fast or slow, not whether it's local or remote [00:19] <Company> well, just use 0-100, where 0 == superfast and 100 = needs avian carrier connection [00:19] <deadChip> yeah true [00:19] <deadChip> the speed is the poitn [00:19] <deadChip> point* [00:20] <mathrick> Company: how do we determine the speed? [00:20] <teuf> deadChip: it depends, if you are interested in file permissions, you don't care if it's slow or fast, but if it's a local fs or a rmeote one :) [00:20] <mathrick> Company: and "net" may matter, as when you are on a dial-up [00:21] <Company> i have no clue how that should be done, i'll leave that to others [00:33] walters (wa...@ve...) joined #gstreamer. [00:41] foser (d0...@22...) left irc: "[ I want to believe ]" [00:44] dolphy (~do...@21...) left irc: "Network down, IP Packets delivered via UPS" [00:48] Nick change: mathrick -> mathrick|away [00:48] abraar (~foo@AToulouse-105-1-1-182.w80-14.abo.wanadoo.fr) left irc: Read error: 110 (Connection timed out) [00:49] ds-uhv (~ds...@ad...) joined #gstreamer. [00:53] pb_ (~pb@2002:5160:45ef:0:240:95ff:fe30:aa84) left irc: "later" [00:57] teuf (~te...@ce...) left irc: Remote closed the connection [00:57] Loevborg (fo...@dy...) joined #gstreamer. [00:57] <Loevborg> ** (process:1188): WARNING **: error setting period in frames (8192): Invalid argument [00:57] <Loevborg> pipeline doesn't want to play [00:58] <Loevborg> what does that mean? (using alsasink on current debian-sid packages) [00:58] <ds> dolphy: zuruck-moo [01:05] Action: sxpert is away: time to go and see dreams [01:05] <Loevborg> noone have a clue? [01:06] <ds> what pipeline? [01:07] <Loevborg> the command that triggered it was "gst-launch sinesrc ! alsasink" [01:07] <ds> alsasink is broken in 0.6 [01:07] <Loevborg> that's a shame! [01:07] <Loevborg> how can I re-configure the output to osssink? [01:08] <ds> uh, sinesrc ! osssink [01:08] hadley_ (~hadley@ACD2EB13.ipt.aol.com) joined #gstreamer. [01:09] hadley (~hadley@ACA7D41D.ipt.aol.com) left irc: Read error: 110 (Connection timed out) [01:09] <Loevborg> that works [01:10] <Loevborg> I have used gstreamer-properties to set osssink as the default output though and now rhythmbox outputs strange noises [01:11] <ds> that should work [01:11] <Loevborg> the sound of the songs is distorted a _lot_ [01:12] <Loevborg> the entire gstreamer avaiable in debian seems to be broken [01:14] <deadChip> rocks... i have bmp playing a file trough gst [01:14] <deadChip> :D [01:14] <ds> Company: you around? [01:24] sublett (~rv...@21...) left irc: "I like food, food is good!" [01:24] <Company> ds: a bit [01:24] Loevborg (fo...@dy...) left irc: "Leaving" [01:31] <ds> I'm trying make sense of gst_structure_from_string() grammar [01:31] <ds> there's a bit of inconsistency [01:32] <ds> i.e., name=value or name:type=value or ... ( type=value, type=value ) [01:32] <ds> while I'm at it, should I fix the grammar, or leave inconsistent? [01:32] <Company> fix it [01:33] <ds> currently, I don't handle types inside lists or ranges [01:33] <Company> and i'd say do it like taaz said [01:33] <ds> ? [01:33] <ds> name = (type) value? [01:34] <Company> stay as close to C structs as possible [01:34] <ds> we are already pretty far away [01:34] <Company> type name = value; type2 name2 = value [01:34] <ds> that's more like declarations [01:35] <ds> in a function [01:35] <Company> yeah whatever [01:35] <ain> why does BBB want me to rewrite wavparse to use rifflib? [01:35] <ain> rifflib blows [01:35] <Company> BBBs new rifflib+ [01:35] <ain> ah [01:35] <ain> but I thought we were doing riffdemux etc? [01:35] Action: Company has no clue [01:36] <ain> seeing as avi and wav etc are so similar [01:36] <ds> ok, I'll come up with some proposals and send it to the list [01:36] Action: ain neither [01:36] <ain> I'll speak to BBB sometime [01:36] ain (~ia...@us...) left irc: "leaving" [01:39] <Company> ds: not specifying a type should be optional and not guaranteed to work [01:39] <Company> ds: apart from that getting all those sets to work might be hard [01:39] <Company> because contents of sets might not have a type [01:39] <Company> or might [01:40] <ds> the only problem I have with "type name = value" is that the type can't be optional [01:40] <Company> why not [01:40] <Company> '=' is not a valid name :) [01:40] <ds> you might want a field name that is the same as a type name [01:40] <ds> ok, maybe it can be optional [01:40] <Company> check if the second string equals '=' [01:41] ain (~ia...@us...) joined #gstreamer. [01:41] <ds> also, for lists, we need a string that means "value of given type [01:41] <ds> like "(type) value" [01:42] <ds> and I [01:42] <ds> (ack!) [01:42] <ds> and I'd like to make that the same as the field assignment, if possible [01:45] <Company> yeah [01:45] <Company> that's true [01:45] <Company> invent something [01:46] <Company> the only thing that has to be guaranteed is that gst_structure_from_string (gst_structure_to_string (x)) == x [01:46] <ds> for a given subset of GTypes [01:47] <Company> yes [01:47] <Company> only if gst_structure_to_string (x) != NULL [01:48] <ds> for any type that converts to/from a string properly [01:53] ain (~ia...@us...) left irc: "leaving" [02:08] seb128 (~seb128@ANancy-111-1-7-76.w81-250.abo.wanadoo.fr) left irc: "I like core dumps" [02:18] <walters> /build/gstreamer-0.7/bin/gst-register-0.7: relocation error: /build/gstreamer-0.7/lib/gstreamer-0.7/libgsttagediting.so: undefined symbol: gst_tag_setter_get_type [02:18] <walters> any ideas? [02:18] <walters> this is with 0.7.2 [02:24] <Company> walters: that should be in the core gst/gsttaginterface.c [02:24] <walters> Company: it's not in 0.7.2 [02:24] <walters> er wait [02:25] thaytan (~ja...@ad...) left irc: Read error: 113 (No route to host) [02:25] <walters> gsttaginterface.c? [02:25] <walters> there's no such file [02:25] <Company> uhm [02:25] <Company> not in 0.7.2, no [02:25] <Company> but in cvs HEAD there should be one [02:25] <walters> Company: sf cvs is borked, otherwise i'd be using head [02:26] <Company> you're obviously using HEAD plugins ;) [02:26] <walters> nope [02:26] <walters> 0.7.2 [02:27] <walters> hm [02:27] <walters> although [02:27] <walters> i have an older build of HEAD in the same dir [02:27] <walters> that's probably the problem. [02:27] <Company> hm [02:27] <Company> maybe [02:27] <Company> the tag interface shouldn't be in 0.7.2 at least [02:28] <Company> though configure.ac shouldn't list that dir [02:28] <walters> yeah, i think i just had leftover plugins from my HEAD build [02:29] <Company> ah [02:29] <Company> your installed plugins, now i understand [02:29] <Company> yeah, that's it probably :) [02:31] Action: Company is going to write a libcaca plugin when coming back home on sunday [02:31] <ds> caca? [02:36] <walters> yay, 0.7.2 works. [02:36] Action: walters now returns you to your regularly scheduled complaining about sf and cvs :) [02:37] leif_ (~le...@ad...) joined #gstreamer. [02:40] <deadChip> how can i query the current play time of a thread? [02:43] <Company> ds: slashdot [02:43] <Company> deadChip: gst_pad_query [02:44] <Company> or gst_element_query [02:44] <deadChip> but what's the property? [02:44] <deadChip> can't find it in the docs [02:44] <Company> look at how gstplay does it [02:44] <deadChip> ok [02:44] <deadChip> thanks [02:44] <Company> plugins/gst-libs/gst/play [02:45] <Company> gnight all [02:45] <deadChip> nite [02:49] leif (~le...@ad...) left irc: Read error: 110 (Connection timed out) [02:49] Company (~Co...@p5...) left irc: Read error: 104 (Connection reset by peer) |