From: <wim...@ch...> - 2001-06-30 04:29:50
|
[06:52] ShrimpX (marius@131.252.244.168) left irc: ircII EPIC4-1.0 -- Are= we there yet? [06:57] ShrimpX (marius@131.252.244.168) joined #gstreamer. [07:20] dap (fU...@po...) left irc:= Read error to dap[pool0417.cvx6-bradley.dialup.earthlink.net]: Connectio= n reset by peer [07:56] steveb (st...@no...) joined #gstreamer. [07:57] Parapraxis (Sp...@us...) left irc: = Client Exiting [08:19] Nick change: taaz -> taazzzz [09:11] Nick change: ajbusy -> ajmitch [09:17] ShrimpX (marius@131.252.244.168) left irc: Read error to ShrimpX[= 131.252.244.168]: Connection reset by peer [09:59] mattsm (ma...@ad...) left ir= c: Client Exiting [10:19] ShrimpX (marius@131.252.244.168) joined #gstreamer. [10:41] mattsm (ma...@ad...) joined = #gstreamer. [10:57] ajmitch (aj...@p6...) left irc: Ping timeout = for ajmitch[p6-max11.dun.ihug.co.nz] [11:11] ajmitch (aj...@p1...) joined #gstreamer. [11:28] Nick change: ajmitch -> ajbusy [12:11] greg_ (gr...@ho...) joined #gstreamer. [13:37] Nick change: ajbusy -> ajzzzz [14:38] greg_ (gr...@ho...) left irc: [x]chat [16:20] dobey (do...@dr...) joined #gstreamer. [16:43] makki (yv...@ma...) joined #gstreamer. [17:35] Parapraxis (Sp...@us...) joined #gs= treamer. [17:49] ChiefHighwater (pa...@te...) joined #gstreamer. [17:49] Parapraxis (Sp...@us...) left irc: = Ping timeout for Parapraxis[user-2ive8ne.dialup.mindspring.com] [18:24] ChiefHighwater (pa...@te...) left irc: Ping timeout f= or ChiefHighwater[temple-baptist.com] [18:27] gtb (syln@Mix-Lyon-203-1-30.abo.wanadoo.fr) joined #gstreamer. [18:27] <gtb> hi! [18:34] Nick change: omega_rr -> omega_ [18:35] <steveb> yo [18:35] <omega_> yo [18:35] <steveb> have you got alsa installed anywhere? [18:36] <omega_> nope [18:38] <steveb> when i run -inspect alsasrc I get warnings like this: Gt= k-WARNING **: gtk_object_arg_get(): could not find argument "ESS_AudioDri= ve_ES1688_0_0_src" in the `h C@' class ancestry [18:38] <steveb> before i delve into the alsa code too much i was wonderi= ng if it might be shim related [18:39] <omega_> no, you want to check the ordering of the arguments on t= he argument registration [18:39] Parapraxis (Sp...@us...) joined #gs= treamer. [18:39] <steveb> the order they are registered in? [18:39] omega__ (om...@bo...) joined #gstream= er. [18:39] <omega__> no, you want to check the ordering of the arguments on = the argument registration [18:39] <omega__> I had a lot of those problems, turns out I just didn't = switch the args around right [18:40] Nick change: omega__ -> omega_afk [18:41] <steveb> are you talking about the args to g_object_class_install= _property and g_param_spec_? [18:42] omega_ (om...@bo...) left irc: Ping t= imeout for omega_[boisdslgw5poolb76.bois.uswest.net] [18:44] gtb (syln@Mix-Lyon-203-1-30.abo.wanadoo.fr) left irc:=20 [18:56] Zeenix (programmer@203.135.10.50) joined #gstreamer. [18:57] <Zeenix> hi [18:58] omega_afk (om...@bo...) left irc: Pin= g timeout for omega_afk[boisdslgw5poolb76.bois.uswest.net] [19:10] ChiefHighwater (pa...@te...) joined #gstreamer. [19:12] <Zeenix> when did that 0.2.1 released [19:13] <ChiefHighwater> Wednesday, i think [19:14] <Zeenix> completely shifted to glib, i think [19:14] <ChiefHighwater> end product was a shift to GObject [19:15] <Zeenix> from GtkObject ? [19:20] <ChiefHighwater> away from, yes [19:20] <ChiefHighwater> but is should shim backwards if GObject isn't th= ere [19:21] <Zeenix> dont know much about Gtk+ hierarchy, i think GtkObject w= as the father of all, right [19:22] <ChiefHighwater> I dunno, ask Wim or Omegafor that one [19:22] <Zeenix> whos Wim, wtay [19:22] <ChiefHighwater> yup, sorry [19:25] <Zeenix> wtay always help, he is a very nice person [19:26] <ChiefHighwater> 8-] when he's awake <g> [19:27] makki (yv...@ma...) left irc: Ping timeout for makk= i[makki.wait-a-bit.com] [19:28] <Zeenix> how to uncompress a bz2 ? [19:29] <Zeenix> s/bz2/.bz2 [19:29] makki (yv...@ma...) joined #gstreamer. [19:31] <ChiefHighwater> Zeenix:I dunno, sorry /me is a windoze user [19:31] Action: ChiefHighwater ducks [19:31] Zeenix (programmer@203.135.10.50) left irc: Read error to Zeenix[= 203.135.10.50]: Connection reset by peer [19:32] Zeenix (programmer@203.135.10.50) joined #gstreamer. [19:36] omega_afk (om...@bo...) joined #gstre= amer. [19:36] Nick change: richardb-asleep -> ricahrdb [19:36] Nick change: ricahrdb -> richardb [19:38] <Zeenix> omega: what that afk means ? [19:39] <omega_afk> away from keyboard [19:39] Nick change: omega_afk -> omega_airport [19:40] <Zeenix> omega: is your keyboard an airport [19:42] taazzzz (dlehn@66.37.66.32) left irc: Ping timeout for taazzzz[66= .37.66.32] [19:42] <omega_airport> no, I'm heading to the airport to go home [19:42] Nick change: wtay-zZz -> wtay [19:42] <wtay> yo [19:42] <omega_airport> be back in 2hrs [19:42] <wtay> heh [19:42] <omega_airport> l8r [19:42] <wtay> have a nice flight :) [19:43] <Zeenix> wtay: hi [19:43] <wtay> yo [19:43] <Zeenix> wtay: I was just telling ChiefHighxxxx about your kindne= ss [19:43] <Zeenix> <g> [19:44] <ChiefHighwater> <Zeenix> wtay always help, he is a very nice pe= rson [19:44] <ChiefHighwater> [10:26] <ChiefHighwater> 8-] when he's awake <g= > [19:44] <ChiefHighwater> [10:27] *** makki (yv...@ma...) Qu= it (Ping timeout [19:44] Action: ChiefHighwater watches wtay blush [19:44] <Zeenix> wtay: what about 0.2.1 [19:44] <wtay> I'm usually asleep :) [19:45] <wtay> Zeenix: yup, a new version [19:45] <Zeenix> a complete shift to GLib [19:45] <wtay> with gtk1.2 compatibility=20 [19:45] <makki> whazzup ? [19:45] <makki> ChiefHighwater: ? [19:46] <wtay> makki: nothing [19:46] <makki> ic [19:47] <Zeenix> wtay: the father of all is now GObject ? [19:48] <wtay> Zeenix: if you compile against glib2 it is, yes [19:50] <Zeenix> wtay: need to update the docs or not ? [19:51] <wtay> Zeenix: what do you mean? [19:51] <Zeenix> wtay: s/docs/gst manual [19:51] <wtay> uhm, yeah.. :( [19:53] <wtay> shouldn't be a lot of work though... [19:53] <Zeenix> i istalled the latest RH Linux yesterday. it contains gt= k+-1.2 [19:53] <richardb> Whee. custom names for element in -launch working. :) [19:53] <wtay> richardb: oh cool! [19:54] <Zeenix> do i need to get glib2 seperatelly [19:54] <richardb> That's about 1/3 of the way to supporting tees... [19:54] <wtay> Zeenix: yes, glib2 is not out yet.. [19:55] <wtay> Zeenix: also, you need the CVS version of gst, I think som= ething is wrong with the gobject compilation [19:56] <richardb> And don't bother trying the current CVS of glib2, that= seems to be broken too...=20 [19:56] <richardb> I made that mistake this morning. [19:56] <wtay> oh [19:56] <wtay> the 1.3.6 tarball then? [19:56] <richardb> v 1.3.6 works fine. [19:56] <richardb> The CVS tag for 1.3.6 seems good, too. [19:57] <wtay> ok [19:57] <Zeenix> wtay: the 0.2.1 tar ball at download.sourceforge wont he= lp=20 [19:57] <Zeenix> ? [19:57] <wtay> Zeenix: nope [19:58] <Zeenix> wtay: dont mind that, but i am getting tired of all this= , i think i should bother of GObject stuff when all is fine [19:58] <wtay> Zeenix: indeed [19:58] <Zeenix> s/should/shouldnt [19:59] <Zeenix> no no no [19:59] <Zeenix> dont have to s/should/shouldnt [20:01] makki (yv...@ma...) left irc: Ping timeout for makk= i[makki.wait-a-bit.com] [20:01] <Zeenix> wtay: but should get the new 0.2.1, right ? [20:02] <wtay> hmm, yes [20:02] <Zeenix> i got the gst tarball, something else needed ? [20:03] <Zeenix> 0.2.1 tar ball [20:03] <wtay> that'll be fine [20:04] <wtay> assuming you have the -devel libs for the other 20+ depend= ecies :-) [20:04] <Zeenix> wtay: i must go home now or i'll die, i am not feeling g= ood today [20:04] <wtay> oh [20:04] <wtay> don't die :) [20:08] makki (yv...@ma...) joined #gstreamer. [20:27] dobey (do...@dr...) got netsplit. [20:28] richardb (ri...@pc...) got netsp= lit. [20:28] Bob (ro...@d1...) got netsplit. [20:28] Parapraxis (Sp...@us...) got netspl= it. [20:28] mattsm (ma...@ad...) got net= split. [20:28] Zeenix (programmer@203.135.10.50) got netsplit. [20:41] _gst_newt_ joined #gstreamer. [21:13] Topic changed on #gstreamer by ChanServ!s@ChanServ: 0.2.1 release= : http://gstreamer.net/releases/0.2.1/ [21:17] PaddyBoy (pa...@ad...) joined #gstreamer. [21:17] <steveb> anyone out there have a built xmms plugin? [21:17] <wtay> I have.. [21:18] <steveb> could you run -inspect on it? [21:18] <wtay> dunno, it doesn't even compile [21:18] <steveb> ok, ta [21:19] <wtay> will try to fix it... [21:20] <steveb> i'm trying to fix alsa [21:25] <wtay> the xmms plugin is seriously broken :( [21:28] <steveb> was it like that pre gobject? [21:28] <wtay> nope [21:36] <makki> xmms plugin ? [21:37] <makki> is that corba communication gstreamer <-> xmms ? [21:37] <wtay> nope, a wrapper around xmms plugins [21:38] mattsm (ma...@ad...) left ir= c: Client Exiting [21:39] <makki> oh, nice [21:39] taazzzz (dlehn@66.37.66.32) joined #gstreamer. [21:39] <wtay> Gtk-WARNING **: gtk_object_arg_get(): could not find argum= ent "XMMS_INPUT_id_software_.cin_player_1.2.5" in the `XMMS_INPUT_id_soft= ware_.cin_player_1.2.5' class ancestry [21:40] <wtay> steveb: is it that what you get? [21:42] <steveb> yay! i'm not going crazy [21:42] <steveb> i'm getting Gtk-WARNING **: gtk_object_arg_get(): could = not find argument "ESS_AudioDrive_ES1688_0_0_src" in the `h C@' class anc= estry [21:42] <Parapraxis> wtay: I've made some changes to gobject2gtk (I think= it should eliminate the compiler warnings ppl were getting), but I'm not= sure how to go about testing the functionality [21:43] <Parapraxis> what's a good way to test it? [21:43] <wtay> Parapraxis: compile an app in test/? [21:43] <Parapraxis> any one? [21:44] <wtay> one with gst_utils_set_arg :-) [21:44] <Parapraxis> I mean test it @ runtime [21:44] <Parapraxis> ok [21:44] <steveb> wtay: alsa, xmms and ladspa are the only plugins with dy= namically generated class names [21:44] <wtay> steveb: the problem is that the new arg names shouldn't ha= ve MyObject:: in front of them [21:45] <steveb> doesn't seem to make a difference for me [21:45] <wtay> that's a first step... [21:45] <steveb> ok, done that then [21:46] <wtay> somthing else is wrong... [21:49] <steveb> wtay: alsa, xmms and ladspa are also the only ones which= don't have their TypeInfo created in a _get_type func [21:49] <wtay> yup [21:50] <steveb> so the GTypeInfo struct isn't const [21:51] <Parapraxis> I just got a simlilar error trying to run test/mp3pl= ay [21:51] <Parapraxis> Gtk-WARNING **: gtk_object_set(): could not find arg= ument "U=E5=EC [21:51] <Parapraxis> = WVS=E8" in the `GstDiskSrc' class ancestry [21:52] <wtay> neat... that used to work though... [21:52] <Parapraxis> that's w/out my gobject2gtk modifications [21:52] <wtay> ugh [21:53] <wtay> I think we need to add NULL to that function/shim... [21:53] <steveb> omega_airport: said that he got a lot of that when the a= rgs which register the params were in the wrong order - but i don't think= that the case with alsa [21:56] <wtay> XMMS_INPUT_id_software_.cin_player_1.2.5: Boolean (Defaul= t true) [21:56] <wtay> XMMS_INPUT_id_software_.cin_player_1.2.5: String (Defaul= t "(null)") [21:56] <wtay> got a little better now.. [21:58] <steveb> how did you do that [21:58] <wtay> GstXmmsInputClass *oclass =3D (GstXmmsInputClass*)(G_OBJEC= T_GET_CLASS(object)); [22:00] <steveb> where? the _set_property func? [22:01] <Parapraxis> does test/mp3play work for either of you? or anythi= ng w/ GstDiskSrc? [22:01] <wtay> everywhere we find the class of the object [22:02] <wtay> Parapraxis: sec... [22:03] <wtay> Parapraxis: mp3play is broke, all the g_object_set functio= ns need a NULL arg at the end [22:04] <Parapraxis> ok [22:04] <wtay> Parapraxis: try aviparse for the gst_util_get* functions [22:05] <wtay> videotest.c:76: warning: passing arg 1 of `gst_util_get_po= inter_arg' from incompatible pointer type [22:05] <wtay> videotest.c:82: warning: passing arg 1 of `gst_util_get_po= inter_arg' from incompatible pointer type [22:05] <wtay> if that's what you're tryin to fix, that is <g> [22:05] <Parapraxis> That's what I did fix, I'm trying to make sure it st= ill runs ;) [22:06] <wtay> uhm.. need something better than aviparse then... [22:06] <wtay> try mpeg2parse3 [22:07] <wtay> ok mp3play works fine here [22:10] <wtay> XMMS_INPUT_xmp_player_2.0.4::active : Boolean (Defau= lt true) [22:10] <wtay> XMMS_INPUT_xmp_player_2.0.4::location : String (Defaul= t "(null)") [22:13] ajzzzz (aj...@p1...) joined #gstreamer. [22:13] Nick change: ajzzzz -> ajmitch [22:23] ShrimpX (marius@131.252.244.168) joined #gstreamer. [22:34] mattsm (ma...@ad...) joined = #gstreamer. [22:34] <wtay> steveb: xmms in CVS has the fixes=20 [22:37] ajmitch (aj...@p1...) left irc: http://www.fr= eedevelopers.net [22:41] makki (yv...@ma...) left irc: Ping timeout for makk= i[makki.wait-a-bit.com] [22:41] steveb (st...@no...) left irc: Ping timeout for ste= veb[node1ee06.a2000.nl] [22:41] steveb (st...@no...) joined #gstreamer. [22:42] <steveb> wtay: thanks [22:44] <Parapraxis> when do elements validate the incoming data (would g= st_element_factory_make ("mpeg2parse",...) fail if the file isn't an mpeg= 2?) [22:45] makki (yv...@ma...) joined #gstreamer. [22:46] <wtay> Parapraxis: _elementfactory_make is just to register the e= lement [22:47] <wtay> validating the data is either done using caps, or in the c= hain function... [22:48] <Parapraxis> ok, any idea why it would be failing? :) [22:48] <Parapraxis> in mpeg2parse3 [22:48] <wtay> oh [22:48] <wtay> lemme see [22:48] <wtay> hmm, it works fine here... [22:51] <Parapraxis> doh, perhaps it isn't being built? [22:52] <Parapraxis> would that give an error, or just silently return NU= LL? [22:53] <wtay> mpeg2parse3 would exit in case of a plugin not found.. [22:54] makki (yv...@ma...) left irc: Ping timeout for makk= i[makki.wait-a-bit.com] [22:55] <Parapraxis> wtay: it does exit, but it doesn't say plugin not fo= und... it just fails an assertion that 'v_decode' (the return from _eleme= ntfactory_make ("mpeg2parse"...)) isn't NULL [22:55] makki (yv...@ma...) joined #gstreamer. [22:56] <wtay> Parapraxis: you most likely don't have the mpeg2dec plugin [22:56] steveb (st...@no...) left irc: [x]chat [22:57] <Parapraxis> ok [23:03] <Parapraxis> man... /me needs a machine about 10x as fast as his = :( [23:07] bln (bj...@h1...) joined #gstreamer. [23:08] <bln> hi [23:08] <wtay> hi [23:12] PaddyBoy (pa...@ad...) left irc:=20 [23:28] Uraeus (csc...@c2...) joined #gstreamer. [23:28] <Uraeus> hi dudes! [23:29] makki (yv...@ma...) got netsplit. [23:29] Uraeus (csc...@c2...) got netsplit. [23:29] richardb (ri...@pc...) got netsp= lit. [23:29] omega_airport (om...@bo...) got netsp= lit. [23:29] ChiefHighwater (pa...@te...) got netsplit. [23:29] wtay (wi...@ca...) got netsplit. [23:29] ShrimpX (marius@131.252.244.168) got netsplit. [23:29] mattsm (ma...@ad...) got net= split. [23:29] Bob (ro...@d1...) got netsplit. [23:29] Parapraxis (Sp...@us...) got netspl= it. [23:30] _gst_newt_ joined #gstreamer. [23:35] <Uraeus> I just saw Antitrust, cool movie! [23:37] <Uraeus> wtay: what can you tell me about festival [23:45] <wtay> hi [23:45] <wtay> sorry, I was afk [23:45] Parapraxis (Sp...@us...) joined #gs= treamer. [23:47] <wtay> Uraeus: festival currently uses a socket connection with t= he festival server to convert text into speech (wav actually) [23:52] ShrimpX_ (marius@Photon.pdx.netroedge.com) joined #gstreamer. [23:54] <Uraeus> wtay: you wrote it? [23:55] <Uraeus> wtay: was it written as a result of the discussions with= the Sun people? [00:00] --- Sat Jun 30 2001 [00:00] bln (bj...@h1...) joined #gstreamer. [00:01] <wtay> Uraeus: I wanted to do it a long time ago, but the discuss= ion helped, yes [00:01] omega_ (om...@om...) joined #gstreamer. [00:01] <wtay> Uraeus: it's nowhere near usable for the sun folks, 'cause= it's way too slow [00:01] <wtay> hi again [00:01] <omega_> yo [00:02] <omega_> the flight was 1+hr late... [00:02] <Uraeus> wtay, omega_: ok, I am updateing my SPEC file and going = to build 0.2.1 RPMS, might even be able to make a lame plugin this time := ) [00:03] <omega_> ok [00:06] <ChiefHighwater> Ello omega_, welcome home 8-] [00:07] <omega_> yo [00:07] <wtay> anyone think sienap will finish the vgasink? [00:07] <Uraeus> wtay: would think so [00:07] <wtay> ok... giving him a few more weeks then <g> [00:07] <Uraeus> hehe [00:08] <Uraeus> wtay: if you are bored, we still need a SID plugin :) [00:08] <wtay> Uraeus: sid is extremely non streamable :) [00:09] <omega_> sid? I didn't think Debian could fit inside of an eleme= nt..... [00:09] <Uraeus> wtay: that is why the task must be undertaken by a maste= r like yourself :) [00:09] Action: wtay slaps omega with a trout [00:09] Kuroyi_ (ri...@ub...) joined #gstrea= mer. [00:09] <wtay> Uraeus: we're doing a "streaming" media framework, remeber [00:10] <Parapraxis> omega_: imagine what would happen if you tried to ad= d the debs to sid [00:12] <Uraeus> wtay: if you don't have the skill for the task there is = no shame in admiting it instead of making excuses <g> [00:12] Action: omega_ hands wtay another trout [00:12] <wtay> Uraeus: ok ok, you convinced me :-) [00:13] <Uraeus> ;) [00:13] Action: Parapraxis strategically locates himself behind omega_ [00:13] <ChiefHighwater> hehe [00:13] <wtay> uhm, sidplay *only* works from a file :-(( [00:14] Apoc (Apoc@212.83.187.119) joined #gstreamer. [00:15] <Parapraxis> UGH! I finally got the mpeg2dec element to be built= and now the next one (colorspace) isn't!!! [00:15] <wtay> ah, maybe it can handle a memory buffer too... [00:15] <Apoc> yo [00:15] <wtay> Parapraxis: libHermes :) [00:15] <wtay> hi [00:16] <Parapraxis> wtay: is that in sid? [00:16] Action: omega_ gets confused now <g> [00:17] <wtay> ii hermes1 1.3.2-2 The He= rmes pixel-format library [00:17] <wtay> ii hermes1-dev 1.3.2-2 Develo= pment libraries for the Hermes pixel-format library [00:17] <Parapraxis> also, wtay or omega_, is the xmlDoc.name ever used? [00:17] <wtay> Parapraxis: not that I know of.. [00:17] <Parapraxis> too hard to grep for [00:17] makki (yv...@ma...) left irc: Back to the real worl= d [00:18] <wtay> Parapraxis: ask the "sedi master" for an appropriate regex= p :-) [00:18] <omega_> um, searching for what? [00:19] <Parapraxis> xmlDoc->name being used [00:19] <omega_> xmlNodePtr.name ? we use that [00:19] <omega_> checking [00:19] <wtay> I don't think we use that... [00:20] <omega_> not afaict [00:20] <Parapraxis> ok [00:20] <omega_> nope [00:21] <omega_> only xmlNodePtr.name [00:21] <omega_> er, ->name [00:22] <Parapraxis> should I leave all function args in the same order a= s in libxml, or clean them up some? (ie xmlSaveFile -> xmlt_doc_write_to= _file should have the doc 1st, but xmlSaveFile has the filename 1st) [00:22] <omega_> be signature compatible with libxml at all cost, imo [00:23] <Parapraxis> hmmm... I would say otherwise, but I'm neurotic abou= t that sort of thing ;) [00:24] <Parapraxis> I was going to do a xmltree -> libxml shim [00:24] <omega_> we should be able to plug in your code in place of libxm= l without doing anything else [00:24] <omega_> xmltree ? [00:24] <Uraeus> wtay: what festival packages do I need? (festival, fring= e & speech_tools) [00:24] <omega_> that's the name of your package? [00:24] <Parapraxis> that's what I called it for now [00:25] <Parapraxis> that's all it does is stick a tree structure over gl= ib's xml parser [00:25] <omega_> what about writing? [00:25] <Parapraxis> it will do that [00:26] <omega_> ok [00:26] <Parapraxis> but that's simple (glib has an xml string escaping f= unction already) [00:26] richardb (ri...@pc...) joined #g= streamer. [00:26] <wtay> hi [00:26] <richardb> hi [00:26] <Parapraxis> hi [00:26] <omega_> well, if there's a shim that has little to no performanc= e impact, then design xmltree however you need to [00:26] <richardb> network outage here. Anything interesting been happen= ing? [00:27] <Parapraxis> omega_: it will all be #defines, some just swapping = argument order [00:27] <omega_> ok [00:27] <wtay> Uraeus: you wabt a SID encoder too? <g> [00:27] <wtay> s/wabt/want [00:28] <omega_> bah. Utah trip cancelled for the moment [00:28] <Parapraxis> I'm not totally convinced it should be a separate pa= ckage... I would think you guys would want to reduce the # of deps gst h= as as much as possible ;) [00:28] <wtay> omega_: how come? [00:28] <omega_> dunno, some scheduling conflict on their end [00:28] <wtay> oh [00:28] <Uraeus> wtay: hmm, well I would think it is impossible to make s= ids out of mp3's or wavs ? [00:29] <omega_> Parapraxis: definitely should be a separate package (ver= y small one) because paranormal will use it, and I can assure you that if= it's really small, it'll be used by a lot of other projects too [00:29] <wtay> Uraeus: just testing.. :-) [00:29] <Uraeus> wtay: but maybe converting a midi to a sid would be poss= ible? [00:29] <omega_> Parapraxis: if you can get it really really small, in co= de space *and* memory requirements, you can get a lot of users [00:29] <wtay> Uraeus: maybe, yes... [00:31] <wtay> Parapraxis: you're doing DOM and SAX? [00:31] <omega_> wtay: heh [00:31] <Uraeus> hmm, is it possible to make a filter which analyzes a mp= 3 or wav and makes a midi file? i would guess it would be next to impossi= ble to have it do it well [00:32] <omega_> Uraeus: almost totally impossible [00:32] <wtay> we might use a sax based parse instead of the current one = to make things faster/smaller... [00:32] <Parapraxis> DOM [00:32] <omega_> wtay: yes, but how on earth would you structure things?? [00:32] <omega_> SAX is callback-based, like the glib parser [00:33] <wtay> omega_: dunno, yet, but it could be done... [00:33] <omega_> wtay: erm, with lots of object callbacks and nastiness, = yes... [00:33] <Parapraxis> wtay: what I'm doing should make it quite small, and= probably a little faster [00:34] <wtay> omega_: there is only one callback function I think=20 [00:34] <omega_> Parapraxis: I threw out 10KB code-size earlier today, yo= u think that's reasonable (stripped, -Os) [00:34] <wtay> omega_: actually very similar to what we do now, but inste= ad of pulling the data out an switching on it we work the other way aroun= d [00:35] <wtay> just an idea... [00:35] <omega_> Parapraxis: any chance of even dropping the glib xml par= ser requirement at some point? [00:36] <Parapraxis> omega_: its possible, but why? [00:36] <omega_> just wondering [00:37] <Uraeus> should I put deinterlace, UDP and chart in their own pac= kages or bundle them with something else? [00:37] <Parapraxis> I'm not doing this in any generalized fashion, it's = just replacing the libxml function calls in gst and pn [00:37] <omega_> right [00:37] <Parapraxis> although they are probably the most-used calls [00:37] <Parapraxis> I'll bbl [00:38] Nick change: Parapraxis -> Para[dinner] [00:38] <wtay> Uraeus: chart and deinterlace could go into one package. U= DP should go into a net package... [00:38] <wtay> Uraeus: or just put them all together, they don't depend o= n anything... [00:39] <Uraeus> wtay: do any of the depend on eachother? [00:39] <Uraeus> s/the/them/ [00:39] <wtay> nope [00:39] <omega_> we need to have a list somewhere of the packages, so we = can derive both the rpms and debs from that list [00:40] <richardb> In gstreamer-launch syntax, what does the comma do in = srcpad names. ie, what's meant by -launch srcelt a,b! ... [00:40] <richardb> ? [00:40] <omega_> to connect several pads in parallel [00:41] <wtay> richardb: for ladspa mainly.. [00:41] <omega_> i.e. filter1 left,right!left,right filter2 [00:41] <richardb> which does filter1.left connected to filter2.left and = filter1.right connected to filter2.right? [00:42] <omega_> yes [00:42] <richardb> Hmm. So how to do filter1.out connected to filter3.le= ft and filter2.out connected to filter3.right?=20 [00:42] Action: richardb ponders [00:43] <omega_> oh, need to have a syntax for *just* connections [00:43] <richardb> filter1[f1] ; filter2[f2] f1.out,f2.out!left,right fil= ter3 [00:43] <richardb> ? [00:44] <omega_> like ... filter1 ; .... filter2 ; filter3 ..... ; filter= 1.out!filter3.left ; filter2.out!filter3.right [00:44] <omega_> maybe ? [00:44] <omega_> though your idea works too [00:44] <richardb> mine is easier to implement right now. :) [00:44] <omega_> yup [00:44] <richardb> working on it already ... though it may be a while... [00:45] Action: wtay hopes nobody is ever going to ask him for a -launch = command for ladspa [00:49] <Apoc> mikmod plugin works ... but with gtkobject :) Somebody kn= ow the next gstreamer release date ;-) [00:49] <Apoc> Please don't say tomorrow ;-) [00:49] <omega_> tomorrow <G> [00:50] <wtay> Apoc: soon.. very soon :) [00:50] <Apoc> Arghh !!! Ok I'll work on it this week end ... [00:51] <omega_> heh [00:54] <Apoc> uraeus : It may be cool to add nickname of people who work= on gstreamer on the status page . Most of them come in #gstreamer ??=20 [00:55] <Uraeus> Apoc: yup, I will add it this weekend, thanks for the su= ggestion [00:56] <Apoc> uraeus : cool ... [01:00] Action: omega_ needs a nap.... [01:03] Action: Uraeus also needs a nap but continues fighting off sleep = in order to save the starving Linux hordes [01:04] <Apoc> I need my bed ;) cya all ! [01:04] <Uraeus> see ya [01:04] <wtay> cya [01:04] Apoc (Apoc@212.83.187.119) left irc:=20 [01:04] <wtay> you're all little wussies :-) [01:06] <wtay> Uraeus: how come gst 0.2.1 is not on gnome.org yet? [01:06] Action: wtay is building a sid plugin [01:06] <Uraeus> wtay: cause I have been to busy to get around to it :(, = but I will add it as soon as I have my RPMS ready for consumption [01:06] <wtay> ok [01:15] <Uraeus> aargh, FLAC has a buggy buildsystem [01:16] <Uraeus> hate it when I have to compensate in my SPEC file for bu= gs in the Makefile setup, almost tempting to learn more about configure a= nd Make in order to be able to fix it myself [01:42] Uraeus (csc...@c2...) left irc: Client Exiting [01:54] gtb (syln@80.9.176.29) joined #gstreamer. [01:54] <gtb> yo! [01:55] <gtb> wtay: why are you planning a reimplementation of CapsNego ? [01:56] <gtb> wtay: what's wrong with the current implementation ? [01:56] <wtay> gtb: the way we handle proxying for example [01:57] <gtb> wtay: only that ? [01:57] <wtay> there's also no way for an element to inspect the peer pad= template [01:57] <bln> anyone know how to check if a GtkWidget reference object ex= ists and are to gtk_widget_destory'ed? [01:57] <bln> s/to/not/ [01:57] <wtay> it's a flag in GtkObject [01:58] <gtb> wtay: a way to imply the application in the negociation pro= cess ? [01:58] <wtay> also, yes [02:06] <wtay> gtb: got ideas to implement a better capsnego? [02:07] <gtb> I began to think on it... [02:07] <gtb> I have writen a little paper describing my tougths [02:07] <wtay> cool [02:08] <gtb> but still nothing very interesting right now [02:08] <wtay> send it to the list? [02:08] <gtb> for the moment, it is written in french, I have to translat= e it [02:09] <gtb> but i'm not sure it is really interesting [02:09] <wtay> I can read french :) [02:09] <gtb> cool !! [02:09] <wtay> un peu <g> [02:10] <gtb> if you want, I can send it to you... [02:10] <wtay> ok [02:10] <gtb> i have just a few things to modify [02:15] <gtb> i've just sent the first part to you [02:16] <gtb> don't expect to find really interesting things, just some t= hougth [02:16] <wtay> I'll read it.. [02:16] <gtb> ok [02:33] ChiefHighwater (pa...@te...) left irc:=20 [02:44] Nick change: taazzzz -> taaz [02:44] <taaz> what's sid? [02:45] <wtay> C64 music :) [02:45] <taaz> uh huh... [02:46] <taaz> add a mikmod plugin while you're at it [02:46] <wtay> someone else is doing that now.. [02:47] <wtay> gtb: you sent the mail? I don't have it yet... [02:47] <wtay> I have to sleep now.. [02:47] <wtay> cya guys [02:47] Nick change: wtay -> wtay-zZz [02:47] Action: taaz almost has 0.2.1 debs done [02:47] <wtay-zZz> cool [02:48] <taaz> 36 packages (!) [02:48] <taaz> almost done with mpeg2dec packaging too [02:49] <gtb> wtay: yes, i send it to u [02:49] <wtay-zZz> gtb: hmm.. I don't have it... wim...@ch...? [02:50] <gtb> wtay: at wt...@ch... [02:50] <wtay-zZz> gtb: uh oh, use wim.taymans instead [02:50] <gtb> wtay: ok, I send it again [02:50] <wtay-zZz> 44 packages upgraded, 2 newly installed, 0 to remove a= nd 0 not upgraded. [02:50] <wtay-zZz> Need to get 40.9MB of archives. After unpacking 17.1MB= will be used. [02:50] <wtay-zZz> ugh [02:51] <gtb> wtay: done [02:53] <gtb> wtay: ask me if you find not clear things [02:56] <wtay-zZz> gtb: ok, gotta sleep now, will read tommorow [02:59] <gtb> wtay: ok, have a good night ! [03:02] Nick change: Para[dinner] -> Parapraxis [03:07] <Parapraxis> omega_: you there? [03:12] <Parapraxis> taaz: is there an mpeg2dec (the lib) package? I cou= ldn't find one today [03:33] <taaz> no, i'm making it [03:34] <taaz> i could be done but the debhelper tools are doing goofy st= uff and not cooperating with me. some automatic stuff isn't. once i fix= that it'll be "done". [03:37] bln (bj...@h1...) left irc: Leaving [03:39] gtb (syln@80.9.176.29) left irc:=20 [03:41] <Parapraxis> omega_: did you want the glib-2 xml parser dropped i= n xmltree to be usable from glib-1.2? [03:43] mattsm (ma...@ad...) joined = #gstreamer. [03:54] <Parapraxis> anyone know how the return values of xmlSaveFile() w= ork? [04:08] <mattsm> Parapraxis: I think it returns the bytes written [04:08] <mattsm> Parapraxis: or -1 if it failed [04:09] <Parapraxis> ok, thanks [05:00] <mattsm> is GstSrc depreciated? [05:00] <omega_> yes [05:01] <omega_> has been for a long long time, actually [05:02] <mattsm> omega_: i am just reading the app developer guide [05:02] <omega_> ok [05:03] <omega_> if you see things like that, file bugs on them [05:03] <omega_> if there isn't a category for that, let me know and I'll= make one [05:03] <mattsm> omega_: well I am reading the old one [05:03] <mattsm> omega_: the new one looks like its what I should be read= ing [05:03] <omega_> yeah, the CVS snapshot [05:07] <Parapraxis> test/mpeg2parse3 segfaults [05:07] <Parapraxis> 0x40832b13 in gst_mpeg2parse_chain (pad=3D0x806b9c8,= buf=3D0x80dc230) [05:07] <Parapraxis> at mpeg2parse.c:984 [05:07] <Parapraxis> 984 guchar byte =3D *(data+offset); [05:07] <Parapraxis> (gdb) print data [05:07] <Parapraxis> $1 =3D (guchar *) 0xffffffff <Address 0xffffffff out= of bounds> [05:07] <Parapraxis> (gdb) print offset [05:07] <Parapraxis> $2 =3D 0 [05:07] <omega_> whoops [05:07] <omega_> what file? [05:07] <Parapraxis> mpeg2parse.c:984 [05:07] <omega_> no, what media file [05:08] <Parapraxis> are there test media files? [05:08] <omega_> gstreamer.net/media/ [05:08] <Parapraxis> oh, ok [05:09] <Parapraxis> ganzo_terminator.mpg [05:09] <omega_> I don't think that's mpeg2 [05:09] <Parapraxis> oh, what is? :) [05:09] <omega_> use bullet.vob or somesuch [05:09] <omega_> that should be mpeg1 video only I believe [05:10] <omega_> we should have an index..... [05:10] <Parapraxis> nm [05:10] <omega_> CLIP04.VOB [05:10] <omega_> thx.vob [05:10] <omega_> those are the two MPEG-2 system streams up there [05:11] <Parapraxis> ok [05:11] <Parapraxis> that's probably the problem [05:11] <omega_> yup [05:11] Action: Parapraxis has no idea what he's doing ;) [05:11] <omega_> however, still file a bug because it shouldn't segfault = I don't think [05:11] <omega_> make it clear that it was parsing MPEG-1 at the time tho= ugh, and which file you were trying [05:11] Action: omega_ starts to wonder how it would fail 'gracefully'...= .. [05:12] <Parapraxis> with a message when it starts up saying 'if this seg= faults, you're using the wrong media format' [05:12] <omega_> heh [05:14] <Parapraxis> ugh... ok, first off, I needed to run -register to m= ake CLIP04.VOB not segfault at start (I'll get the hang of this sooner or= later) [05:14] <omega_> file a bug on that, too [05:14] <Parapraxis> secondly, it segfaulted when the clip was done [05:14] <omega_> yeah, known problem, we're solving that next week <g> [05:15] <Parapraxis> lol... perhaps you should hold off on solving the se= gfaults-when-invalid-media problem and work on the segfaults-when-valid-m= edia problem ;P [05:16] <omega_> heh [05:16] <Parapraxis> btw, how much error checking are you expecting from = xmltree? [05:16] <omega_> um, dunno [05:16] <omega_> I have no idea what the choices are, when it comes to an= xml parser/writer [05:17] <Parapraxis> neither do I :( [05:17] <omega_> doh [05:17] <omega_> ask DV [05:17] <Parapraxis> DV? [05:17] <omega_> Daniel Veillard [05:17] <omega_> (sp?) [05:18] <Parapraxis> is correct ;) [05:18] <omega_> ok [05:19] <omega_> if you do talk to him, make it very clear why you're wri= ting a new parser/writer <g> [05:19] <omega_> else he'll probably squish you like a bug ;-) [05:20] <Parapraxis> Actually, I mean doing things like checking that a n= amespace is actually above a node when creating a new node (if the ns is = being set) [05:20] <omega_> um, I dunno, do whatever is easy for now, we can break i= t later... [05:21] <omega_> you can trivially solve the problem by documenting the p= otential issues, with a big =02DON'T DO THAT!=02 attached [05:21] <omega_> it is a specialized parser after all [05:23] <Parapraxis> it's so specialized, that it's not even a parser ;) [05:24] <Parapraxis> omega_: any reason why you didn't add my gst/gstcpu.= c changes? [05:24] <omega_> haven't gotten around to it yet [05:27] <Parapraxis> oh, I can do it if you like; the only externally vis= ible change is that edx isn't output [05:27] <omega_> ok, if you want to, go for it [05:34] <omega_> grmbl [05:35] Action: omega_ goes and looks at bitstream_show_bh_mmx() again an= d figures out what's wrong with it [05:35] <omega_> ah, loaded a dword into the wrong register, fixed [05:37] <omega_> hmmm, I wonder... [05:37] <omega_> I have a 64 entry table that contains the result of 64 -= <index> [05:37] <omega_> it's dwords, so movd can do its thing [05:37] <omega_> but it's 8 cache lines... [05:38] <omega_> should I just do the math instead? [05:38] <Parapraxis> but your not crossing cache lines [05:38] <omega_> so? [05:38] <omega_> that's .25KB out of 16KB [05:38] <Parapraxis> what kinda math is it? [05:38] <omega_> non-trivial amount [05:39] <omega_> just 64 - numbits [05:39] <Parapraxis> just do the math [05:39] <omega_> maybe use leal ? [05:39] <omega_> the cache-line polution is what worries me the most [05:39] <Parapraxis> so time it ;) [05:40] <omega_> I can't generate a usable workload with cache usage atm [05:40] <Parapraxis> 64 - numbits is the math? [05:40] <omega_> yes [05:40] <Parapraxis> how is that non-trivial? [05:40] <omega_> intel uses a lookup in their getbits too [05:40] <omega_> non-trivial amount of memory wasted, cache-lines chewed = up [05:40] <Parapraxis> ah [05:41] <Parapraxis> 1 sub =3D 1 cached mem read [05:41] <omega_> leal maybe [05:41] <omega_> checking docs [05:41] <Parapraxis> I think leal is for muls [05:41] <omega_> powers of two, yes <g> [05:42] <omega_> but it does an other math too, within certain limits [05:42] <Parapraxis> but sub is 1 uop [05:42] <omega_> true [05:42] <omega_> I'll make it a macro and see how it times with no cache-= pressure [05:44] <omega_> we'll see how smart gcc is when I stick a "r"(64-numbits= ) in the constraints [05:45] Action: Parapraxis has lost faith in gcc being smart about constr= aints [05:45] <omega_> heh [05:46] <omega_> movl $64, %eax [05:46] <omega_> subl 8(%edx),%ecx [05:46] <omega_> um, wait [05:47] <omega_> where's the subl I want? [05:47] <Parapraxis> hehe [05:48] <omega_> ergm#&#$@ [05:48] <omega_> movl $64, %eax [05:48] <omega_> movl %eax %esi [05:49] <omega_> subl %ebx, %esi [05:49] <omega_> ... movd %esd, %mm4 [05:49] <omega_> ok, that's sad [05:50] <omega_> let's see how the top-half fares [05:51] <mattsm> omega_: does gstreamer do mime-type detection from files= ? [05:51] <omega_> by way of the typefind element, yes [05:51] <mattsm> hrmm [05:52] <mattsm> let me try again [05:52] <omega_> it might be useful for a source to pick up at least a mi= me/type from the file extension, if possible [05:52] <mattsm> can an autoplugger contruct a pipeline from a disksrc an= d a osssink? [05:52] <omega_> but we can't assume that'll happen, so we use the typefi= nd instead [05:52] <omega_> yes [05:52] <omega_> autoplugger uses typefind internally [05:52] <mattsm> omega_: ok, what functions am I looking for? [05:53] <omega_> gst/autoplugger/gstautoplugger.c [05:53] <omega_> er, autoplug/ [05:54] Action: omega_ thinks something like an __attribute__ ((outline))= would be nice to have on certain basic blocks associated with if's [05:55] <omega_> Parapraxis: do you know about objdump -S ? [05:55] <Parapraxis> what about it? [05:55] <omega_> try it on one of your .o's [05:56] <Parapraxis> didn't know that, that's nice [05:56] <omega_> yu[ [05:56] <omega_> er, yup [05:57] <omega_> it's not that accurate though, it completely leaves out = some of the code I see in the .S [05:57] <Parapraxis> yeah, I first tried it on my xmltree.o & had 0 outpu= t [05:58] <omega_> hmm [05:58] <omega_> it doesn't seem to decode the entire call to the bottom = half... [05:58] <omega_> trying just -D [05:59] <omega_> um [06:00] <omega_> there it is [06:00] <omega_> odd [06:00] <omega_> I had to recompile [06:02] <omega_> the code gcc is generating is certainly non-optimal [06:04] Action: Parapraxis wonders how much of a speedup he could get fro= m taking over register management in his blur routine [06:04] <omega_> depends... [06:05] <Parapraxis> actually, none, I think [06:05] <omega_> should thorougly study the generated asm and see what if= anything is done stupidly [06:05] <Parapraxis> I do the loop in asm [06:05] <omega_> in my case I have some C at the front, and am currently = relying on gcc to be smart about certain constraints [06:06] <omega_> the strategy is failing, and I think I know how to both = fix it, *and* outline the code all at once [06:07] <omega_> I'm also tempted to use __attribute__ (regparm(2)) for m= y bottom half [06:07] <omega_> and/or go so far as to not even call it , but to jump to= it [06:08] <omega_> but I have a feeling that the standard C function stuff = will get in the way then, and I'd have to write the bottom halves as a .S= instead [06:08] <Parapraxis> you're calling the function in a tight loop? [06:08] <omega_> unless regparm does what I think it does [06:08] <omega_> no, all over the place [06:08] <omega_> see http://codecs.org/projects/ewocc/mpeg4.parse [06:08] <omega_> every single symbol:integer there turns into symbol =3D = bitstream_get(bs,integer) [06:09] <omega_> and/or a vlc codeword decode, esp for dct coeffs, is tou= chy [06:11] <Parapraxis> hmmm [06:13] <omega_> hrm, /me is an idiot [06:13] <omega_> I left an old register constraint in there, maybe that w= as making gcc have a cow? [06:14] <omega_> still doing a mov $64,%eax;sub %edx,%eax [06:14] <Parapraxis> is ewocc actually working? I didn't realize what it= was exactly... that's quite cool ;) [06:14] <omega_> but at least it's not shuffling it through %esi [06:15] <omega_> yeah, there's a semi-functioning version in CVS [06:15] <omega_> and I wrote a perl pre-processor yesterday that can gene= rate an almost fully working mpeg2 bitstream parser [06:15] <omega_> but I really need to get a grammar functioning [06:16] <omega_> the current ewocc.c is a completely hand-written parser = though, and thus it sux [06:17] <Parapraxis> I thought you were using lex for that [06:17] <omega_> no, I used the glib lexer for that one [06:18] <Parapraxis> blech [06:18] <omega_> hey, it worked <g> [06:18] Action: Parapraxis doesn't like the glib lexer very much [06:18] <Parapraxis> <omega_> the current ewocc.c is a completely hand-wr= itten parser though, =02and thus it sux=02 [06:19] <omega_> yeah? [06:19] <omega_> lexer + parser + ... =3D compiler [06:19] <omega_> flex + bison + .... =3D compiler [06:20] <omega_> glib + =02ewocc.c=02 =3D who knows?? [06:20] <Parapraxis> not I [06:20] Action: omega_ does: a mess [06:20] <Parapraxis> heh [06:20] ChiefHighwater (fl...@su...) joined #gst= reamer. [06:20] <omega_> scan through that code, you'll see what I mean [06:20] <omega_> yo [06:20] <ChiefHighwater> Ello [06:21] <Parapraxis> helL [06:21] <omega_> s/$/o/ [06:22] <Parapraxis> if he can take a letter off the end, why can't I? [06:22] Action: Parapraxis is getting too tired [06:22] <omega_> erm <g> [06:23] <omega_> Parapraxis: ok, I subtract one reg from another, I want = to jump only if the result is < 0 [06:23] <omega_> what insn do I use? [06:23] <Parapraxis> jl [06:23] <Parapraxis> I think [06:24] <omega_> hmm, makes sense <g> [06:24] <Parapraxis> is there a jn [06:24] <Parapraxis> I think they're the same, but my thinking isn't quit= e up to par right now [06:24] <omega_> heh [06:24] <Parapraxis> and jn seems safer [06:25] Action: omega_ can't find a reference for what the flags really m= ean.... [06:26] <Parapraxis> which ones? [06:27] <omega_> SF, OF [06:27] <omega_> jl jumps if SF !=3D OF [06:27] <Parapraxis> look at the sub instr [06:28] <omega_> neither subl or cmpl pages say what happens other than "= set according to the result" [06:28] <Parapraxis> hold on, I have a book ;) |