IRC Logs

*******************************************************************
[03:01] Action: wheels is away: sleep
[03:05] aldug ([email protected]) joined #gstreamer.
[03:06] Miko5881 ([email protected]) joined #gstreamer.
[03:07] <Company> "... File a bug."
[03:08] Action: ds considers that as his new "I'm back" message
[03:09] <Company> was my mail too bad considering thomasvs and me are good at arguing with each other?
[03:12] <ds> can't tell, I've been drinking and/or smoking
[03:15] <jimmy_dean> Company: why would a gboolean element member assignment survive from one function to another but none of my pointer memeber assignments?  All of them are coming back NUL
[03:15] <jimmy_dean> Company: for example, I have this assignment at the bottom of my init function: equalizer->db_table = db_table;
[03:15] <jimmy_dean> Company: db_table is a double *db_table
[03:16] <jimmy_dean> Company: and so is equalizer->db_table
[03:16] <Company> then both point to the same memory
[03:16] <Company> the question is: what did you do with the memory they point to?
[03:17] <jimmy_dean> Company: I did this: db_table = malloc(1000 * sizeof(double));
[03:17] <jimmy_dean> Company: and looped through assigning values which I can verify are getting set
[03:18] <Company> and then you do an equalizer->db_table = db_table
[03:18] <jimmy_dean> right
[03:18] <Company> and after that equalizer->db_table[0] should be equal to db_table[0]
[03:18] <jimmy_dean> Company: then in my chain function, try and do this: double *db_table = equalizer->db_table;
[03:18] <jimmy_dean> Company: right, they should be
[03:19] <jimmy_dean> but then in my chain function both are NULL
[03:19] <Company> where do you set that?
[03:19] <Company> in what init function
[03:19] <jimmy_dean> Company: you just gave me an idea...and I was right
[03:19] <jimmy_dean> Company: I found my error
[03:20] <Company> great
[03:20] <jimmy_dean> Company: thanks, this solves it all
[03:21] <Company> ffcolorspace 565=>other_cspace is broken
[03:23] <ds> file a bug
[03:24] i10neorg ([email protected]) joined #gstreamer.
[03:25] <Company> i'm about to
[03:30] hadley ([email protected]) joined #gstreamer.
[03:32] i10neorg ([email protected]) left #gstreamer.
[03:38] Action: ds declares his new laptop "OK"
[03:40] desrt ([email protected]) left irc: Client Quit
[03:40] desrt ([email protected]) joined #gstreamer.
[03:41] Miko5881 ([email protected]) left irc: Remote closed the connection
[03:42] <thaytoo> ds: what did you buy?
[03:43] <Company> hopefully a 4bit machine ;)
[03:43] <Company> and a 64bit one
[03:43] <ds> emachines m5312
[03:43] <ds> I would have gotten the opteron model, but they don't have it yet
[03:45] aldug ([email protected]) left irc: Success
[03:46] Action: wheels notes that there's not much point in getting an Opteron at this point.  They're slower than the fastest Athlons and ~EUR 100 more...
[03:47] Action: Company notes that 64bit is a good reason for a dev to get one, especially if he writes data decoders
[03:48] <Company> but that model has shared video memory...
[03:48] <Company> and 3 month aol!
[03:50] <thaytoo> ds: let me know what the battery life is like
[03:50] <wheels> Company: Well, if you're writing decoders your bottlenecks are still most likely on getting that data to and from the CPU (i.e. disks, networks).  Encoders is a different story.  :-)
[03:50] <thaytoo> everyone keeps asking me for comparisons and I don't know any :)
[03:51] Action: wheels works on Opterons a lot at work...
[03:51] <Company> wheels: but you want to have your code endian and CPU width clean
[03:51] <wheels> Company: Opteron won't help on endianness...
[03:52] <Company> i'm not obsessed with fast but with correct
[03:52] <Company> yeah, but powerbooks are expensive and not 64bit either
[03:52] Action: wheels just has a shell account on a PA-RISC machine that he uses for testing such things.  :-)
[03:53] <Company> i don't
[03:53] <Company> and i hate remote debugging via DSL...
[03:54] <wheels> Company: If you want to try a handful of archs it's really the only way...
[03:54] <wheels> ...or you spend a whole lot on machines.
[03:54] <wheels> ...or you do kernel work and then people send them to you.  :-)
[03:55] <Company> or i just buy an ibook and an opteron :)
[03:56] <ds> thaytoo: need to get 2.6 running first
[03:56] <khalek> or a sparc64 box?
[03:56] <ds> thaytoo: the fan is on continually, which is not a good sign
[03:57] <khalek> they aren't overly expensive second hand
[03:57] <ds> hopefully I'll be getting a dual Opteron soon
[03:57] <Company> sparcs aren't nice
[03:58] <Company> ibooks are great laptops and opterons are fast laptops
[03:58] <Company> sparcs are just cluttering your room
[03:59] <Company> and they're probably slow which is bad for compiling and testing
[03:59] <khalek> sure I have an ibook, but I wouldn't mind a sparc64 as well :)
[03:59] <khalek> think they require alignment or am I thinking of something else
[04:00] <ds> feng-shui?
[04:13] ChriHJW ([email protected]) left irc: Read error: 110 (Connection timed out)
[04:32] pb_ ([email protected]) left irc: "Client exiting"
[04:56] <jimmy_dean> Has anyone talked to Steve before who is the author of the LADSPA plugins?
[04:57] <jimmy_dean> Steve Harris is his full name
[05:08] somex1 ([email protected]) joined #gstreamer.
[05:33] harshy ([email protected]) joined #gstreamer.
[05:57] sri ([email protected]) left irc: "Client exiting"
[06:06] harshy ([email protected]) left irc: "I quit for now"
[06:17] Nick change: bluejay|gone -> bluejay
[06:19] sri ([email protected]) joined #gstreamer.
[06:19] <sri> y0
[06:19] <sri> anyone around?
[06:19] <sri> question on the mixer.
[06:19] <sri> and also gst-mad.c breaks..and I don't know how to fix it quite yet
[06:20] ChriHJW_log ([email protected]) joined #gstreamer.
[06:20] <thaytoo> sri: I'm not
[06:21] <thaytoo> what's wrong?
[06:23] <sri> never mind fixed my gstmad.c prob
[06:23] <sri> but mixer is broken
[06:23] <sri> running gst-mixer gives me:
[06:23] <sri> parsing: "endianness = (int) BYTE_ORDER, signed = (boolean) { TRUE, FALSE }, width = (int) { 8, 16 }, depth = (int) { 8, 16 }, rate = (int) [ 1000, 48000 ], channels = (int) [ 1, 2 ]"
[06:23] <sri> returned 0 "(int) BYTE_ORDER, signed = (boolean) { TRUE, FALSE }, width = (int) { 8, 16 }, depth = (int) { 8, 16 }, rate = (int) [ 1000, 48000 ], channels = (int) [ 1, 2 ]"
[06:23] <sri>  
[06:23] <sri> (gst-mixer:32342): GStreamer-CRITICAL **: Could not convert static caps "audio/x-raw-int, endianness = (int) BYTE_ORDER, signed = (boolean) { TRUE, FALSE }, width = (int) { 8, 16 }, depth = (int) { 8, 16 }, rate = (int) [ 1000, 48000 ], channels = (int) [ 1, 2 ]"
[06:23] <sri> Segmentation fault
[06:23] <sri> I have no idea about this.
[06:36] ChrisHJW_log ([email protected]) left irc: Connection timed out
[06:38] <thaytoo> sri: I'm not sure when the 'BYTE_ORDER' constant was added
[06:38] <thaytoo> possibly after whichever gstreamer version you have
[06:43] <sri> hmm.
[06:43] <sri> mineis from anoncvs@freedesktop
[06:44] <thaytoo> well, it'll be in there for sure
[06:44] <sri> any ideas on waht to do here?
[06:44] <sri> my regstry file has like 4 lines in it
[06:45] <sri> unfortunately, none of my gui mixers are working thanks to this thing
[06:51] <thaytoo> well, the caps look ok to me
[06:51] <thaytoo> so I'm not sure :/
[06:52] hadley ([email protected]) left irc: "Leaving"
[06:52] <sri> let me delete the system registry and rebuild
[06:53] <sri> maybe some old plugins or something causing problems.
[06:55] <thaytoo> I'm heading home from work
[06:55] <thaytoo> back later
[06:55] thaytoo ([email protected]) left irc: "leaving"
[08:18] pgoel ([email protected]) left irc: Read error: 113 (No route to host)
[08:18] pgoel ([email protected]) joined #gstreamer.
[08:32] Shoragan ([email protected]) joined #gstreamer.
[08:34] thaytan ([email protected]) joined #gstreamer.
[08:38] jowenn ([email protected]) joined #gstreamer.
[08:39] <jowenn> hi
[08:42] pgoel ([email protected]) left irc: Read error: 113 (No route to host)
[08:44] pgoel ([email protected]) joined #gstreamer.
[08:44] rk ([email protected]) joined #gstreamer.
[08:45] <rk> morning
[08:46] walters ([email protected]) left irc: "out"
[08:55] jowenn ([email protected]) left #gstreamer.
[08:59] kmaraas ([email protected]) left irc: "Leaving"
[09:09] neiras ([email protected]) joined #gstreamer.
[09:21] foser ([email protected]) joined #gstreamer.
[09:32] swentel ([email protected]) joined #gstreamer.
[09:34] neiras ([email protected]) left irc: "Download Gaim: http://gaim.sourceforge.net/"
[09:37] somex1 ([email protected]) left irc: Remote closed the connection
[09:44] Nick change: bluejay -> bluejay|sleep
[09:44] sub_pop ([email protected]) left irc: "Client exiting"
[09:55] Nick change: mathrick|sleep -> mathrick
[09:55] <mathrick> mornin
[10:08] <thaytan> yo
[10:11] Action: ds remembers to check stuff in before sleep
[10:39] thomasvs ([email protected]) joined #gstreamer.
[10:40] thomasvs ([email protected]) left irc: Remote closed the connection
[10:40] dolphy ([email protected]) joined #gstreamer.
[10:41] harshy ([email protected]) joined #gstreamer.
[10:41] <dolphy> morning
[10:44] thomasvs ([email protected]) joined #gstreamer.
[10:44] markov ([email protected]) joined #gstreamer.
[10:47] <thomasvs> mornaroo
[10:48] <mathrick> yo thomasvs, markov, dolphy, harshy :)
[10:48] <harshy> lo
[10:50] alley_cat ([email protected]) joined #gstreamer.
[10:50] <mathrick> lo alley_cat
[10:55] <alley_cat> morning
[10:55] <ds> thomasvs: could you fix the directory 'basement' in CVS?
[10:55] <ds> it's not writeable
[10:56] <thomasvs> ok
[10:56] <dolphy> thomasvs: please also take a look to the fixes i had to do yesterday to get the whole thing to build
[10:56] <thomasvs> dolphy: in cvs ?
[10:56] <dolphy> thomasvs: GST_LIBRARY_ERROR_ENCODE added to gsterror.h
[10:56] <dolphy> thomasvs: for xvidenc
[10:56] <thomasvs> ah
[10:56] <thomasvs> testing code you can't compile is hard :)
[10:56] <dolphy> thomasvs: and prototype of _gst_element_error_printf before the macro in gstelement.h
[10:57] <ds> dolphy: I fixed the switch pipeline you gave me yesterday by implementing gst_value_serialize()
[10:57] <thomasvs> ds: done - try it
[10:57] <ds> so now caps are serialized in a completely locale independent way
[10:57] Action: thomasvs goes through Company's remarks
[10:57] <thomasvs> ds: great
[10:58] <dolphy> ds: i fixed switch yesterday too :)
[10:58] Action: thomasvs is going to look at media-info today
[10:58] <dolphy> ds: implementing proxying of link
[10:58] <ds> dolphy: well, there was that problem, too
[10:58] <dolphy> ds: and forcing src caps to sink requested pads
[10:58] <dolphy> ds: it now works like a charm
[10:59] <dolphy> ds: i was able yesterday to have 30 seconds of the video signal and then 30 seconds of visualisation and then goto first one
[10:59] <dolphy> ds: reading matrix.avi 
[10:59] <dolphy> ds: was fun but too 99% of CPU :)
[11:00] Nick change: harshy -> harshyZ
[11:00] <ds> dolphy: really, you should be unlinking and re-linking to the video stream you want
[11:01] <ds> otherwise, both video and viz will take cpu time
[11:01] <dolphy> ds: yeah so switch is useless then
[11:02] <dolphy> ds: i realized that yesterday when i understood that i will have to add/remove the visualization bin anyway each time
[11:02] <dolphy> ds: so no need for switch
[11:02] <dolphy> ds: sucks..
[11:03] <thomasvs> well, it's a good element to have around :)
[11:03] Action: thomasvs is happy dolphy fell into the classic trap
[11:04] <dolphy> hehe :)
[11:04] <dolphy> you like when i loose my time, don't you ? :)
[11:05] <thomasvs> heh, no I like when you contribute to the pool of plugins
[11:06] <dolphy> well yeah switch might be usefull to someone someday.. :-)
[11:06] <dolphy> who knows
[11:06] Rotty ([email protected]) joined #gstreamer.
[11:10] Company ([email protected]) left irc: Remote closed the connection
[11:13] wheels ([email protected]) left irc: "using sirc version 2.211+KSIRC/1.3.9"
[11:18] ross_ ([email protected]) joined #gstreamer.
[11:18] Nick change: ross_ -> rburton
[11:19] <rburton> can i set tags with gst-launch-0.7?
[11:34] markov ([email protected]) left irc: "bbl"
[12:09] sxpert_work ([email protected]) got netsplit.
[12:11] sxpert_work ([email protected]) returned to #gstreamer.
[12:11] #gstreamer: mode change '+b *!*omg@*.client2.attbi.com' by leguin.freenode.net
[12:32] hyriand ([email protected]) left irc: "p00f"
[12:42] sublett ([email protected]) joined #gstreamer.
[12:53] thomasvs ([email protected]) left irc: Read error: 110 (Connection timed out)
[12:54] <thaytan> http://jaime.hemmett.org/gallery/LCA2004/imgp3950 <- photo from LCA, steve preparing for his talk
[12:57] dolphy ([email protected]) left irc: Read error: 110 (Connection timed out)
[12:58] thomasvs ([email protected]) joined #gstreamer.
[13:07] Action: thomasvs tries to figure out why mad always settles on layer=1
[13:09] sublett ([email protected]) left irc: Read error: 113 (No route to host)
[13:11] BBB ([email protected]) joined #gstreamer.
[13:12] sublett ([email protected]) joined #gstreamer.
[13:12] <BBB> howdy all
[13:13] <thaytan> yo yo
[13:13] <thomasvs> so, I made vorbisfile create a tag with stream properties.  Does anyone know how it gets signalled higher up ?
[13:14] <thaytan> there's a have_tag signal somewhere, I think?
[13:15] <thomasvs> yeah, but it's not in the elements, so I'm assuming it needs to be done through a function
[13:15] <thomasvs> the whole tag API is pretty dense
[13:15] LeRoutier ([email protected]) joined #gstreamer.
[13:16] <LeRoutier> hello
[13:16] <thomasvs> hm, found_tags maybe
[13:17] <BBB> found_tags is the bin signal
[13:17] <BBB> I think for tags, you need to implement the taginterface
[13:17] <BBB> and then you get the have_tag signal
[13:18] <BBB> emit that, and the tagging stuff will take care that the containing bin gets signalled
[13:23] Company ([email protected]) joined #gstreamer.
[13:29] <BBB> hi company
[13:31] Action: rburton wonders if BBB or Company knows how to set tags using gst-launch
[13:31] <Company> rburton: not possible
[13:31] <rburton> :(
[13:31] <BBB> I was just gonna say the same
[13:31] <Company> good morning angels btw
[13:31] <BBB> feel free to implement it ;)
[13:31] <BBB> good mroning charlie
[13:31] <Company> please not
[13:31] <BBB> -typos
[13:31] <Company> gst-launch was always supposed to be simple...
[13:31] <BBB> heh, gst-launch as full-featured replacement of commandline tools like mencoder or ffmpe
[13:31] <BBB> :p
[13:32] <Company> though you could write a settag element and i wouldn'
[13:32] <Company> t complain
[13:32] <rburton> i wanted a quick and easy way to put tags in a file :)
[13:32] Action: BBB likes that
[13:32] <thomasvs> Company: I wanted to throw an error from tag_add_valist, but it doesn't have a pointer to the element
[13:32] <BBB> rburton: how about a small python script?
[13:32] <Company> src ! settag title=title ! encoder ! sink
[13:32] <thomasvs> Company: should it have something like that ?
[13:32] <Company> thomasvs: no
[13:33] <rburton> BBB: i guess this is as good a time as any to try the python bindings
[13:33] <Company> thomasvs: because tags are supposed to be used by apps, too
[13:33] <BBB> rburton: well, dunno, I didn't test the tagging yet either
[13:33] <rburton> heh
[13:33] <BBB> I'll implement tagging in wav, avi & matroska as I make progress
[13:33] <BBB> and port wavenc/parse to rifflib
[13:34] <thomasvs> Company: right - so should it have a GError or GstError then ?
[13:34] <BBB> GError, I guess
[13:34] <BBB> gerror is pretty much standard
[13:34] <LeRoutier> current CVS HEAD : gstdivxenc.c:300:50: macro "gst_element_error" requires 5 arguments, but only 4 given
[13:34] <thomasvs> yeah - but then it should probably print instead
[13:35] <Company> why would that throw an error anyway?
[13:35] <BBB> LeRoutier: will fix in a sec.
[13:35] <LeRoutier> k
[13:36] Action: BBB updates CVS and rebuilds
[13:36] hyriand ([email protected]) joined #gstreamer.
[13:37] <Company> thomasvs: there's no reason why add_valist would need a GError
[13:38] <thomasvs> Company: an app/plugin trying to set a tag that's not registered
[13:38] <Company> thomasvs: that's a programming error
[13:39] <Company> it uses g_return_if_fail for a reason there :)
[13:39] <thomasvs> right, and I disagree that hte programming error should be done this way, since it's nonobvious
[13:39] <thomasvs> a lot of errors we throw now are programming errors
[13:40] <thomasvs> ie, all of the "couldn't negotiate caps" errors are programming errors for example
[13:40] <Company> no, they're not
[13:40] <Company> a programmer cannot know valid caps
[13:40] <Company> or if valid caps exist at all
[13:41] <Company> but a programmer can know if a tag is registered yet
[13:41] <thomasvs> how can a programmer know if it's registered ?
[13:41] dolphy ([email protected]) joined #gstreamer.
[13:42] <BBB> hi dolphy
[13:42] <Company> thomasvs: gst_tag_exists
[13:45] <BBB> LeRoutier: fixed
[13:45] <BBB> oh crap, someone else edited the file
[13:45] Action: BBB kicks people that commit stuff that he's editing
[13:45] <LeRoutier> BBB: thomas fixed it at the same time 
[13:46] <BBB> well, there were some typos left
[13:47] <BBB> ...
[13:47] Action: BBB fixes
[13:47] <dolphy> re
[13:47] <LeRoutier> lut dolphy
[13:49] sublett ([email protected]) left irc: "I like food, food is good!"
[13:51] <thaytan> there seems to be something unitialised in xvimagesink
[13:51] <thaytan> or I've missed some logic
[13:52] <thaytan> randomly, my playback window will have a window border, and sometimes now
[13:52] <thaytan> not
[13:52] <Company> and sometimes after resizing the window is hidden
[13:52] <Company> and there's only the black background
[13:52] <thaytan> oh, I haven't found that one
[13:53] <Company> i think that happens in gst-player mopre often than with a direct window
[13:53] <thaytan> what are you putting in gstui ?
[13:54] <thaytan> I've been wondering about a nasty developer's test element that creates a gtkwindow with some seek buttons in it that I can whack in a pipeline
[13:54] <thaytan> probably would interact badly with threading, but I don't tend to use many threads
[13:55] <thaytan> anyway, I think I will send a patch with this subtitle stuff to the list
[13:55] <thaytan> along with my lengthy todo list
[13:58] <LeRoutier> divx[enc/dec] fixed. next problem :
[13:58] <LeRoutier> gstmpeg2enc.cc: In function `void gst_mpeg2enc_loop(GstElement*)':
[13:58] <LeRoutier> gstmpeg2enc.cc:242: error: parse error before `__null'
[13:58] <LeRoutier> gstmpeg2enc.cc:254: error: parse error before `__null'
[13:58] <LeRoutier> make[3]: *** [libgstmpeg2enc_la-gstmpeg2enc.lo] Erreur 1
[13:58] <thomasvs> LeRoutier: maybe you should send in patches
[13:59] <BBB> LeRoutier: I'm working on it
[13:59] <LeRoutier> thomasvs, you really don't want it, my C is very bad and i don't know gstreamer internals
[14:00] <BBB> LeRoutier: I'll be there in a sec
[14:00] <thomasvs> BBB: too late
[14:00] Action: BBB is currently configuring plugins
[14:00] <BBB> hum
[14:00] Action: BBB kicks thomas
[14:00] <BBB> thomasvs: will you fix dxr3 too then?
[14:00] <thomasvs> BBB: no, if you want to go ahead
[14:00] <thomasvs> I can't compile it
[14:01] <BBB> you can't compile mpeg2enc either
[14:01] <BBB> :p
[14:01] <thomasvs> I know :)
[14:01] <BBB> tsk
[14:01] <BBB> :D
[14:01] <BBB> nayway
[14:01] <BBB> MPEG time for me
[14:01] <BBB> bbias
[14:03] <dolphy> thaytan: don't know i ll take a look at the window manager thing 
[14:12] iain ([email protected]) joined #gstreamer.
[14:16] Company ([email protected]) left irc: Remote closed the connection
[14:22] <thaytan> okay, dvdnav stuff for comment on its way to the list
[14:24] <BBB> cool!
[14:25] <thaytan> don't get too excited :)
[14:25] <thaytan> loads to do yet
[14:25] <thaytan> I need feedback
[14:28] <BBB> I'll try to commentr
[14:28] <BBB> loads to do here, too
[14:28] <BBB> ;)
[14:28] <thaytan> ta 
[14:28] <thaytan> g'night for now :)
[14:28] <LeRoutier> BBB: mpeg fixed but the same type of error there :
[14:28] <LeRoutier> gstmplex.cc: In function `void gst_mplex_loop(GstElement*)':
[14:28] <LeRoutier> gstmplex.cc:282: error: parse error before `__null'
[14:28] <LeRoutier> make[3]: *** [libgstmplex_la-gstmplex.lo] Erreur 1
[14:28] <LeRoutier> good night thaytan
[14:31] <BBB> #ifdef ENABLE_NLS
[14:31] <BBB>   setlocale (LC_ALL, "");
[14:31] <BBB>   bindtextdomain (GETTEXT_PACKAGE, LOCALEDIR);
[14:31] <BBB>   textdomain (GETTEXT_PACKAGE);
[14:31] <BBB> #endif /* ENABLE_NLS */
[14:31] <BBB> that's in dxr3's plugin
[14:31] <BBB> shouldn't that be gone?
[14:32] <thomasvs> why should it be gone ?
[14:33] <BBB> oh, other plugins have it too
[14:33] <BBB> nevermind
[14:34] <BBB> no, I thought the orig author put it there and it dodn't beloing there
[14:34] <BBB> it was just a missing include
[14:34] <thomasvs> yeah, just add the i18n include
[14:34] <BBB> indeed
[14:34] <BBB> LeRoutier: I'm on it
[14:34] <BBB> unless thomas beats me
[14:35] <BBB> thomasvs: oh, btw, g++ seems to require (NULL) instead of NULL
[14:35] <thomasvs> BBB: yep, but make it ("")
[14:35] <thomasvs> BBB: I'm going to change all of them today too
[14:35] <BBB> I made of all them NULL for now
[14:35] <thomasvs> hm
[14:35] <thomasvs> (NULL) should be ok, IIRC
[14:36] <BBB> uhm
[14:36] <BBB> I just made it "
[14:36] Action: BBB won't tocuh it again
[14:37] <thomasvs> no, make it (NULL) for consistency
[14:40] <thomasvs> hm
[14:40] <thomasvs> BBB: can you get tags from a vorbis file ?
[14:42] <thomasvs> hmph
[14:42] Action: thomasvs doubts we can get tags from vorbis at all right now
[14:46] mathrick ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[14:48] sjoerd ([email protected]) left irc: Remote closed the connection
[14:51] sublett ([email protected]) joined #gstreamer.
[14:53] <BBB> not sure, thomasvs, I'll try
[14:54] <BBB> (how do I test?)
[15:00] <BBB> thomasvs: you forgot to cvs add gsttypefind.[ch], I'm affraid
[15:00] <thomasvs> BBB: gst-launch -t filesrc ... ! vorbisfile
[15:00] <thomasvs> BBB: don't think so, try doing an update
[15:00] <thomasvs> BBB: I did it in the cvsroot directly
[15:00] <BBB> Removed files:    gst/elements    : gsttypefindelement.c gsttypefindelement.h
[15:01] <BBB> oh, right
[15:01] <thomasvs> check first, ask later :)
[15:01] <BBB> then the commit msg is bad
[15:01] <thomasvs> it's part of
[15:01] <thomasvs> the other part can't be logged :)
[15:01] <BBB> (hey, I only read commit msgs, I never check out directly)
[15:01] <thomasvs> BBB: no, but if you do an update, you'll see it's ok
[15:01] <thomasvs> ok, so
[15:01] <thomasvs> the gpl headers
[15:01] <thomasvs> when should you add your name ?
[15:01] <thomasvs> and when not ?
[15:02] <BBB> when you wrote large parts of it
[15:02] <BBB> (imo)
[15:02] <BBB> or if you maintain it
[15:02] <BBB> maybe s/or/and/
[15:02] <BBB> you retain copyright either way
[15:02] <thomasvs> application/vnd.rn-realmedia: ra, ram, rm
[15:02] <BBB> so that doesn't matter
[15:02] <thomasvs> this looks wrong no ?
[15:03] <BBB> I think that's the official mimetype
[15:05] <thomasvs> ugly batshit
[15:05] <LeRoutier> hey thomasvs and BBB, you know what ?  gstreamer/gst-plugins now compiles well. seems all i18n and error handling changes are ok now
[15:07] <LeRoutier> so, congrats guys
[15:07] <thomasvs> thanks
[15:07] <thomasvs> waiting for the next one to have some more libs to complain :)
[15:07] sjoerd ([email protected]) joined #gstreamer.
[15:07] <LeRoutier> i have really little plugins that are disabled
[15:08] <BBB> seems to all compile for me now, too
[15:09] <LeRoutier> loaded 135 plugins with 277 features. only ones i don't have are : arts*, dxr3, gsm* and a few others
[15:09] <BBB> thomasvs: yes, the mimetype is ugly, but I didn't mak eit up myself
[15:09] Action: BBB has dxr3
[15:09] Action: BBB has 304 features
[15:09] <BBB> all compiles now ):
[15:09] <BBB> :)
[15:10] <BBB> loaded 148 plugins with 309 features
[15:10] <thomasvs> BBB: so, does vorbis work for you ?
[15:13] Action: dolphy only has 264 :)
[15:16] <BBB> gst-launch filesrc location=/media/cds/Deftones\ -\ Deftones/9\ -\ Bloody\ Cape.ogg ! vorbisfile
[15:16] <BBB>  runs
[15:16] <BBB> how do I extract metadata?
[15:16] <thomasvs> -t
[15:16] <thomasvs> like I said
[15:16] <thomasvs> :)
[15:16] <BBB> oh, -t
[15:16] <BBB> I missed that flag
[15:17] <BBB> well, I don't get any output
[15:17] <thomasvs> me neither
[15:17] <thomasvs> crap crap crap
[15:17] <thomasvs> this is going to take some work
[15:17] <BBB> it does have tags, though
[15:18] <thomasvs> and, of course, now I have no way at all to distinguish between metadata tags and streaminfo, for example
[15:18] <thomasvs> this completely blows
[15:18] Action: thomasvs thinks about making GstTag have a type
[15:19] <BBB> can't you make it a subclass of tag?
[15:19] <BBB> GstStreamInfo extends GstTag
[15:19] <BBB> (no more)
[15:20] <thomasvs> I don't know
[15:20] <thomasvs> but the theory behind the new way is dumber than what we had before
[15:21] <thomasvs> it's pretty much useless to keep the features
[15:21] <thomasvs> and I'm not sure benjamin understands what it's about
[15:21] Action: thomasvs writes a mail
[15:21] <thomasvs> or, I would if I could get to my mail server
[15:22] <swentel> heh
[15:22] <swentel> bash wiebe :)
[15:22] <thomasvs> swentel: I'm trying to get to him, but he's not answering
[15:22] <swentel> or maybe the new site got hacked <g>
[15:23] <thomasvs> wouldn't be surprised, damnit
[15:26] sjoerd ([email protected]) left irc: Remote closed the connection
[15:27] <BBB> thomasvs: I think benjamin is trying to do what he feels is best... what's wrong with the new system?
[15:27] <BBB> it has some nifty features, like forwarding in pipelines
[15:27] <BBB> filtered caps as a way of setting tags is pretty cool imo (theory)
[15:28] <BBB> or well, something like that
[15:28] <thomasvs> there are three types of "metadata"
[15:28] <thomasvs> one kind describes the audio
[15:28] <thomasvs> ie, title, artist, ...
[15:28] <thomasvs> the things you can change without touching the data
[15:28] <thomasvs> the second kind is the encoding metadata
[15:29] <thomasvs> what codec is used, bitrate, ...
[15:29] <thomasvs> the kinds of things you can only change by transcoding the encoded format
[15:29] <thomasvs> the third kind is the metadata describing parameters of the decoded digital audio
[15:29] <thomasvs> rate, channels, ...
[15:29] <thomasvs> these three are different both in function, ways of changing them, ways of being attached ot the data stream, ...
[15:29] <thomasvs> and are also different in how they should be displayed
[15:29] <BBB> for the app, they should be the same
[15:29] <thomasvs> right now, benjamin only implemented the first kind
[15:30] sjoerd ([email protected]) joined #gstreamer.
[15:30] <thomasvs> BBB: no they shouldn't.  you can't change sample rate the way you can change title.
[15:30] <BBB> I think, actually, that (2) and (3) would be 'technical'
[15:30] <BBB> I'm not sure if it's totally the same
[15:30] <thomasvs> if what is totally the same ?
[15:31] <BBB> my idea is that (2) should be part of GstCaps
[15:31] <thomasvs> BBB: right, that's what was in the plugins
[15:31] <BBB> but maybe that's a bit weird
[15:31] <thomasvs> BBB: but benjamin ripped it out
[15:31] <thomasvs> BBB: also, benjamin does have a bitrate tag
[15:31] <BBB> no, I mean as part of capsnego
[15:31] <thomasvs> so I have no idea where he was going
[15:31] <thomasvs> BBB: what, bitrate ?
[15:31] rburton ([email protected]) left irc: "Client exiting"
[15:31] <BBB> the srccaps should give bitrate if the element knows
[15:31] <BBB> yes
[15:31] Rotty ([email protected]) left irc: Remote closed the connection
[15:31] <thomasvs> why would that have anything to do with caps ???
[15:31] <BBB> and similar things
[15:31] <thomasvs> the bitrate has nothing to do with caps
[15:32] <BBB> it does
[15:32] <thomasvs> elements can handle stuff regardless of bitrate
[15:32] <BBB> no
[15:32] <thomasvs> not to mention the fact that they can be variable
[15:32] <BBB> asf, mpeg (as example) require the bitrate
[15:32] <BBB> and if its variable, they need to know the average bitrate
[15:32] <thomasvs> why do they need that ?
[15:32] <BBB> avi needs it too for audio
[15:32] <BBB> mpeg needs it for seeking
[15:32] <BBB> asf/mpeg both need it to determine stream speed if we're doing streaming video
[15:33] <BBB> avi needs it for seeking (note that avi has no other way than this to seek audio in the file)
[15:33] <thomasvs> BBB: the mpeg element knows the bitrate anyway ?
[15:33] <BBB> how?
[15:33] <BBB> by simply pre-reading it? that's not single-pass
[15:34] <thomasvs> BBB: I mean, why should it need to expose it at all ?
[15:34] <thomasvs> BBB: any of the mpeg elements knows the bitrate by seeing the header
[15:34] <thomasvs> BBB: where else do they get the bitrate from ?
[15:34] <BBB> I'm talking about muxers
[15:34] <BBB> if I have videosource ! mpegencoder ! mpegmuxer, mpegencoder needs to tell the muxer the chosen bitrate
[15:35] <thomasvs> BBB: I'm talking about decoders.  for encoders it's another thing.
[15:35] <BBB> or the resulting mpeg file won't play well at all in players (e.g. WMP is quite picky), and not at all on hw players
[15:35] <BBB> decoders/parsers should read it in the same way
[15:35] <BBB> it should be part of the caps
[15:35] <BBB> and you could use that in streaminfo
[15:36] <thomasvs> BBB: the end result would be caps renego *all the time* when vbr is used
[15:36] <BBB> no, *average bitrate*
[15:36] <BBB> not current bitrate
[15:36] <BBB> ;)
[15:37] <thomasvs> BBB: so you would signal current bitrate how ?
[15:37] <BBB> why do you need current bitrate?
[15:37] <thomasvs> because apps want it ?
[15:38] <BBB> do they want current bitrate for every second in the video?
[15:38] <BBB> how is that interesting?
[15:40] Nick change: khalek -> kh_zZz
[15:48] <thomasvs> BBB: start xmms, play a vbr file
[15:48] <thomasvs> BBB: it's not about what we think it's interesting, it's about what people want from a framework
[15:49] Rotty ([email protected]) joined #gstreamer.
[15:51] <BBB> thomasvs: hm... I doubt that's interesting, but well, your choice
[15:56] <thomasvs> BBB: since a) xmms shows it and b) xmms is a typical example of an app we want to be a framework for it therefore is of no relevance if we think it's interesting
[16:04] teuf ([email protected]) joined #gstreamer.
[16:05] <teuf> hi
[16:05] Rotty ([email protected]) left irc: Read error: 113 (No route to host)
[16:08] Rotty ([email protected]) joined #gstreamer.
[16:29] Company ([email protected]) joined #gstreamer.
[16:33] <LeRoutier> seems latest typefind changes broke gstreamer/po compilation
[16:33] Company ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[16:33] <teuf> LeRoutier: works for me
[16:33] <teuf> (with a cvs update from about 30 minutes ago)
[16:33] <LeRoutier> teuf, compiling from a clean CVS tree ?
[16:34] <teuf> I didn't run make clean in my tree, but rerun autogen
[16:35] <thomasvs> LeRoutier: do an update
[16:35] <thomasvs> LeRoutier: what are you seeing specifically ?
[16:36] <LeRoutier> gentoo CVS ebuilds have one big disavantage : they are slow but a bigger advantage : they never compile directly from the local CVS repo. they copy it somewhere for compilation
[16:36] <LeRoutier> w8, looking at it
[16:36] <LeRoutier> basicaly, no rule to make gst...typefind...c ...
[16:37] <thomasvs> ah, ok
[16:37] <LeRoutier> Making all in po
[16:37] <LeRoutier> make[2]: Entering directory `/usr/portage/tmp/portage/gstreamer-0.7.3/work/gstreamer/po'
[16:37] <LeRoutier> make gstreamer-0.7.pot-update
[16:37] <LeRoutier> make[3]: Entering directory `/usr/portage/tmp/portage/gstreamer-0.7.3/work/gstreamer/po'
[16:37] <LeRoutier> make[3]: *** Pas de règle pour fabriquer la cible « ../gst/elements/gsttypefindelement.c », nécessaire pour « gstreamer-0.7.pot-update ». Arrêt.
[16:37] <thomasvs> fixing it
[16:37] Rotty ([email protected]) left irc: Read error: 113 (No route to host)
[16:37] <LeRoutier> gentoo CVS ebuilds have one big disavantage : they are slow but a bigger advantage : they never compile directly from the local CVS repo. they copy it somewhere for compilation
[16:38] <thomasvs> why is htat better ?
[16:38] <thomasvs> personally, I  think gentoo shouldn't do cvs builds
[16:38] <LeRoutier> because it catches problems like that that people not making make clean/autogen at each time wont catch
[16:38] <LeRoutier>  
[16:39] <LeRoutier> well, gentoo did not do a CVS ebuild for gstreamer. i did one for me
[16:39] <thomasvs> LeRoutier: yeah, but I don't see why that's better :)
[16:39] <thomasvs> we run a testbuild for that three times a day
[16:39] <teuf> LeRoutier: I prefer to get work done rather than having to wait for a full rebuild each time ;)
[16:39] <LeRoutier> yep, true that gst compilation takes time
[16:40] <LeRoutier> aroudn 12 mins for gst-plugins here
[16:42] <LeRoutier>  
[16:42] <LeRoutier> does gstreamer has a CVS activity stat page or a CVS IRC bot ? (i doubt for the bot)
[16:43] <teuf> LeRoutier: _gst_newt_ is a bot afaik
[16:43] <teuf> LeRoutier: and there is a mailing list with daily irc logs iirc
[16:43] <thomasvs> LeRoutier: #commits on freenode logs us
[16:44] yippi ([email protected]) joined #gstreamer.
[16:44] <teuf> hm, I missed the CVS part of your message LeRoutier ;)
[16:45] <LeRoutier> oh, k. i didn't know it went to freenode #commit
[16:46] hadley ([email protected]) joined #gstreamer.
[16:46] Action: BBB is becoming a real MPEG master
[16:46] <LeRoutier> my project uses the CIA bot which posts commit summaries to the project chan and maintains a web page with commits. (look here if you don't know it : http://cia.navi.cx/stats/project/prelude)
[16:47] <LeRoutier> it's cool. (even something could be cool and useless at the same time)
[16:47] <LeRoutier> +if
[16:48] <LeRoutier> it also maintains a RSS feed of commits
[16:49] <BBB> planet.gnome.org should have CVS commit summaries per day
[16:49] Company ([email protected]) joined #gstreamer.
[16:51] Rotty ([email protected]) joined #gstreamer.
[16:53] <LeRoutier> http://cia.navi.cx/stats/project/GStreamer
[16:53] <LeRoutier> seems it already has stats for gstreamer
[16:59] tlg ([email protected]) joined #gstreamer.
[17:00] swentel ([email protected]) left irc: 
[17:14] Rotty_ ([email protected]) joined #gstreamer.
[17:14] Rotty ([email protected]) left irc: Read error: 113 (No route to host)
[17:19] Nick change: LeRoutier -> LeRaway
[17:32] <BBB> thomasvs: does it work again?
[17:32] Action: BBB checks commits
[17:32] Action: BBB doesn't want to try it out ;)
[17:32] Action: BBB is lazy
[17:33] cleith ([email protected]) joined #gstreamer.
[17:33] <thomasvs> BBB: it ?
[17:33] <cleith> Hello
[17:33] <thomasvs> hi
[17:33] <cleith> Anyone here use pigeon?  Or can you direct me to a place to ask about pigeon?
[17:34] <thomasvs> pigeon ?
[17:34] <thomasvs> do you know what channel you're in ?
[17:34] <cleith> cd burner (requires gstreamer)
[17:34] <thomasvs> I haven't heird of it yet
[17:34] <BBB> thomasvs: vorbis streaminfo+tags
[17:35] <thomasvs> heard
[17:35] <thomasvs> BBB: streaminfo yes, tags no
[17:35] <cleith> https://sourceforge.net/projects/gnome-cd-burner/
[17:35] <thomasvs> BBB: I'm trying to send a mail about it but my mail server is acting up
[17:35] <BBB> pigeon? probably coaster?
[17:35] <thomasvs> cleith: haven't used it yet.  any screenshots available ?
[17:35] <BBB> I think oggdemux reads tags, no?
[17:36] <thomasvs> BBB: it doesn't look like it no
[17:36] <BBB> nope
[17:36] <BBB> indeed
[17:36] Action: BBB will fix avi some day
[17:36] <cleith> thomasvs, hmmm, not sure
[17:36] <BBB> time to go home, bye
[17:36] BBB ([email protected]) left irc: "Client exiting"
[17:37] <cleith> thomasvs, http://gnome-cd-burner.sourceforge.net/ for screenshots
[17:42] jowenn ([email protected]) joined #gstreamer.
[17:42] <jowenn> hi
[17:43] <thomasvs> hi jowenn
[17:43] <thomasvs> cleith: ah, now I remember seeing it before
[17:43] <thomasvs> cleith: does it work ?
[17:44] <cleith> Well, I think so, for others that is (it has had several releases now), but it does not work for me
[17:44] <thomasvs> cleith: and you think it's gst related ?
[17:44] <cleith> the package I have (rpm) depends on gstreamer >= 0.6.4
[17:45] ChriHJW_log ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[17:45] <thomasvs> cleith: oh, so it's a packaging problem ? and you cannot install it ?
[17:46] <cleith> no it installs fine, and seems to work right up until it you actually ask it to burn the cd, then it eventually hangs.
[17:47] <thomasvs> but no idea why it hangs ?
[17:48] <jowenn> Is it allowed to post 9 lines of code here ? I have a question about how to solve that problem of those lines in a clean way
[17:48] Nick change: LeRaway -> LeRoutier
[17:49] <thomasvs> jowenn: for once, yeah
[17:49] <jowenn> while (gst_bin_iterate(GST_BIN(pipeline))) {
[17:49] <jowenn> ....
[17:49] <jowenn>         if (someCondition) {
[17:49] <cleith> thaytan, no, the cd drive even spins as if it attempts to burn the cd, but, nodda
[17:49] <jowenn>                 GstElement *src=gst_bin_get_by_name(GST_BIN(pipeline),"source1");
[17:49] <jowenn>                 gst_element_set_state(src, GST_STATE_NULL);
[17:49] <jowenn>                 g_object_set(G_OBJECT(src),"location", "mynewfile.mp3", NULL);
[17:49] <jowenn>                 gst_element_set_state(src, GST_STATE_PLAYING);
[17:49] <jowenn>        }
[17:49] <jowenn> }
[17:49] <LeRoutier> isn't iain working on pigeon ?
[17:50] <thomasvs> LeRoutier: you would htink, from the font, but IIRC, no
[17:50] <thomasvs> jowenn: ok, and the question ?
[17:50] <jowenn> I want to dynamically replace one of the sources of a pipeline while the other sources should continue without interruption
[17:50] <jowenn> the above works, but it produces  warning about bad timestamps
[17:51] <thomasvs> jowenn: hm.  not sure from meory if it should theoretically work this way.
[17:51] <thomasvs> but I see what you want.
[17:51] <jowenn> It doesn't work though if I put the source to a sink of an adder and the source had an EOS before
[17:51] <jowenn> s/put/connect/
[17:52] <thomasvs> jowenn: ok, so the src in your case is which exactly ?
[17:52] hadley ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[17:53] <jowenn> I have three filesrc objects, with files of different length. After the shortes one has finished playing, I want to start another file or restart the same file
[17:53] <jowenn> those filesrc objects are connected to mad plugins, which are connected to an adder
[17:54] <jowenn> the src with name "source1" in the example above is a filesrc
[17:55] <thomasvs> ok, so the easiest thing to get what you want is to not reuse input pipes
[17:56] <thomasvs> when a new file starts, you just connect a new bin with a new reader and decoder to adder
[17:56] <Company> it's supposed to work this way
[17:56] <Company> but it currently doesn'T because there's is no proper event wrapping of streams
[17:58] <jowenn> thomasvs: and the old / finished one can be removed, or would that cause problems ?
[17:59] <cleith> Anyone here use coaster?
[17:59] Company ([email protected]) left irc: Read error: 113 (No route to host)
[18:00] Company ([email protected]) joined #gstreamer.
[18:00] <thomasvs> jowenn: it can be removed, yeah
[18:00] <thomasvs> cleith: can it be used eyt ?
[18:04] <cleith> i can not get pigeon to work 100% yet, no.  What do you use to burn cd's?
[18:05] <thomasvs> I haven't burnt any in a long time
[18:05] <jowenn> thanks
[18:05] Nick change: harshyZ -> harshy
[18:06] rk ([email protected]) left #gstreamer ("heading home").
[18:10] <thomasvs> Company: if I know a GstCaps struct is at a given memory address, how do I print it in gdb ?
[18:11] Nick change: harshy -> harshyWork
[18:11] <cleith> Later
[18:11] cleith ([email protected]) left #gstreamer ("Leaving").
[18:11] <jowenn> cu
[18:12] jowenn ([email protected]) left irc: "using sirc version 2.211+KSIRC/1.3.9"
[18:19] eliman_ ([email protected]) joined #gstreamer.
[18:20] chef-de-police ([email protected]) joined #gstreamer.
[18:21] eliman_ ([email protected]) left #gstreamer ("Leaving").
[18:24] chef-de-police ([email protected]) left irc: "Leaving"
[18:31] mxpxpod ([email protected]) joined #gstreamer.
[18:35] markov ([email protected]) joined #gstreamer.
[18:36] apoc ([email protected]) joined #gstreamer.
[18:37] <apoc> hi
[18:38] teuf ([email protected]) left irc: "Client exiting"
[18:59] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[19:01] apoc_ ([email protected]) joined #gstreamer.
[19:07] <Company> thomasvs: print gst_caps_to_string ((GstCaps *) 0xdeadbeef)
[19:19] apoc ([email protected]) left irc: Read error: 110 (Connection timed out)
[19:23] mathrick ([email protected]) joined #gstreamer.
[19:27] Nick change: apoc_ -> apox
[19:27] Nick change: apox -> apoc
[19:27] mxpxpod ([email protected]) left irc: Remote closed the connection
[19:29] <mathrick> re
[19:29] tlg ([email protected]) left irc: "Client exiting"
[19:38] <thomasvs> going home, later
[19:49] thomasvs ([email protected]) left irc: Read error: 60 (Operation timed out)
[19:51] Nick change: LeRoutier -> LeRaway
[19:59] Misirlou ([email protected]) joined #gstreamer.
[20:17] <Misirlou> Company: ping
[20:22] mathrick_ ([email protected]) joined #gstreamer.
[20:22] mathrick_ ([email protected]) left irc: Client Quit
[20:36] thomasvs ([email protected]) joined #gstreamer.
[20:47] manu ([email protected]) joined #gstreamer.
[20:47] dolphy ([email protected]) joined #gstreamer.
[20:48] <dolphy> re
[20:52] <ds-work> boing
[20:54] Action: thomasvs just moved 500 MB of mail between servers
[20:55] hallibaby ([email protected]) joined #gstreamer.
[20:56] manu ([email protected]) left #gstreamer ("A plus tard").
[20:58] Nick change: bluejay|sleep -> bluejay
[20:59] <dolphy> ds-work: http://getcontrol.sourceforge.net/files/gstplay_pipeline.png
[20:59] <dolphy> ds-work: what about that pipeline for gstplay then
[20:59] <dolphy> ds-work: with replugging 
[21:01] <ds-work> you want to disconnect both tee ! visualization and visualization ! queue
[21:02] <dolphy> ds-work: just disconnecting visualization ! queue won't work ?
[21:03] <dolphy> ds-work: the visualization element has to be removed from the pipeline when disconnected ?
[21:06] <mathrick> dolphy: won't tee block on push to inactive peer?
[21:07] <mathrick> dolphy: also, IIRC, there were some issues wrt. to EOS signalling with unconnected elems, so it looks like you'd need to remove that part, not only disconnect it
[21:08] <mathrick> s/to EOS/EOS/
[21:08] <dolphy> mathrick: yeah i have played with that one already
[21:09] <mathrick> dolphy: that one == ?
[21:10] <dolphy> mathrick: having unconnected element preventing EOS
[21:11] <mathrick> dolphy: k, and what was the solution? is it resolved in general, or you did some hack in your pipeline?
[21:13] <dolphy> mathrick: well that was a programming error
[21:13] <dolphy> mathrick: it's normal that EOS is not triggered if an element is not connected
[21:13] <dolphy> mathrick: so you should not have an orphan element in a bin if you want to receive EOS
[21:14] <mathrick> dolphy: so you've solve it in your pipeline?
[21:14] <mathrick> s/solve/solved/
[21:14] <ds-work> dolphy: no, because the visualization element(s) will still get the audio, so will still use CPU
[21:15] <mathrick> dolphy: if not, one possible hack I see is to install custom EOS handler on the queue after vis/vid, and have it set vis part to eos on eos arriving to sink
[21:15] <ds-work> ugh, could someone remove John Fleck from planet.gnome.org?
[21:15] <ds-work> it's like reading slashdot
[21:16] <mathrick> ds-work: why, you mean b/c it's planet.GNOME.org?
[21:16] <ds-work> yeah
[21:16] <ds-work> because I have yet to see any gnome-related post
[21:17] <mathrick> ds-work: i remember seeing one or two, but it's been a while
[21:19] <mathrick> ds-work: as to gstplay pipeline, what will be harder / more hackish: removing / adding vis part each time, or forwarding eos from queue to vis part?
[21:20] <ds-work> what does forwarding eos matter?
[21:20] <mathrick> ds-work: huh? could you rephrase?
[21:21] <ds-work> I didn't understand your question
[21:21] <mathrick> ds-work: <mathrick> dolphy: if not, one possible hack I see is to install custom EOS handler on the queue after vis/vid, and have it set vis part to eos on eos arriving to sink
[21:21] <mathrick> ds-work: it's to avoid non eos'd vis part blocking eos on pipeline, btw.
[21:22] <mathrick> s/non/not/
[21:22] <dolphy> mathrick: i really prefer adding the vis element to the pipeline, connect it and play again
[21:22] <dolphy> mathrick: that save cpu
[21:22] <ds-work> yeah, linking/unlinking is best
[21:22] <mathrick> dolphy: if you disconnect it on both ends, it doesn't suck cpu too
[21:22] <mathrick> s/too/either/
[21:22] <ds-work> I'm not sure the pipeline will run with a disconnected element in PAUSED
[21:23] <ds-work> might as well remove it
[21:25] <dolphy> ds-work: yeah so basically when settting visualization i pause the pipeline if vis_element has peers on pads (mean it is inside) unlink, remove it and add the new one
[21:26] <ds-work> yes
[21:28] Nick change: Jara[zZ] -> Jaramir
[21:28] kmaraas ([email protected]) joined #gstreamer.
[21:28] <mathrick> hmm, come to think of it, is there any way for for ex. player using gstplay, or even just spider, to easy control 'postprocessing' option of decoder (for ones supporting it)? IMO it should be easily accessible, as times when I had only PII@400 aren't that far away, and watching anything with postprocessing > 0 was impossible, yet OTOH I have Athlon 2k now, and would like to have some postproc done
[21:31] <ds-work> what do you mean by postprocessing?
[21:33] <dolphy> ds-work: gst_element_unlink (tee, element) works even if tee has multiple pads ?
[21:34] <mathrick> ds-work: things like MPEG4 motion compensation something, image smoothing and such. They're nice cpu suckers, but enhance quality
[21:35] <mathrick> ds-work: for ex, DS DivX filter has slider labeled "postprocessing" with possible levels 0-5, which somehow alters that
[21:38] <ds-work> dolphy: it should.  otherwise it's a bug
[21:38] <ds-work> mathrick: we should probably have a 'quality' interface for that
[21:41] <dolphy> ds-work: and that might destroy the requested pad
[21:41] <dolphy> ds-work: using gst_pad_unlink is probably safer if i want to keep that requested pad
[21:42] <mathrick> ds-work: hmm, ok, looks like quality is well fitted for that. But at least mpeg4 decoders tend to have various options that aren't easily translated into quality sliders, like hw iDCT support, and all sorts of other params (don't really know how much thery're relevant for gst, seen them on DS filters), and these should IMO also be exposed somehow
[21:43] <ds-work> not really
[21:43] <ds-work> mathrick: anything that's not 'quality' should not be exposed to a user
[21:44] <ds-work> dolphy: yeah, probably
[21:45] <mathrick> ds-work: hmm, wouldn't it be a little bit limiting for power users (where "users" == apps of any sort)?
[21:46] <ds-work> mathrick: would those "power users" be the ones who like to fiddle with controls even though they don't do anything useful?
[21:46] <mathrick> ds-work: the thing is, are we really able to set automagically all parameters that may be of interest / influence on app functioning? Something like pro video editor may demand that sort of functionality
[21:47] <ds-work> elements still have properties
[21:47] <mathrick> ds-work: but how do you access elems inside spider?
[21:48] <ds-work> spider is just a bin
[21:49] <mathrick> ds-work: hmm, if that's how it is, then ok. IIRC, Company said spider is not-to-be-touched, so I thought it wasn't an option
[22:00] thaytan ([email protected]) left irc: "See y'all"
[22:10] Uraeus ([email protected]) joined #gstreamer.
[22:10] <Uraeus> gæday
[22:12] <mathrick> yo Uraeus
[22:13] <apoc> hey Uraeus
[22:13] <dolphy> Uraeus: hey :)
[22:16] Action: Uraeus spent one hour one the phone today arguing with an ignorant co-worker who claimed UTF-8 is the same as US ASCII, and blaming the use of UTF-8 in the database for norwegian characters not working in his report
[22:17] <ds-work> heh
[22:17] <Uraeus> that everthing else had them was 'just a coincidence'.....argh
[22:17] <dolphy> ds-work: how can i check why gst_element_link returns false ?
[22:20] <ds-work> dolphy: I think it's always due to not finding compatible pads
[22:20] <sxpert> Uraeus: hahahhahaa
[22:20] <sxpert> Uraeus: friggin' moron
[22:22] <dolphy> ds-work: yeah just found that in gst-debug
[22:22] Nick change: LeRaway -> LeRoutier
[22:22] <dolphy> ds-work: for any reason it tells that goom can not connect to queue
[22:22] <dolphy> ds-work: probably because queue is still connected to something else
[22:22] <dolphy> ds-work: probably a threading issue, right ?
[22:23] <dolphy> ds-work: i m ghosting video queue to video thread
[22:24] <dolphy> ds-work: and unlinking autoplugger from video_thread
[22:24] <dolphy> ds-work: and then try to connect goom to video_thread
[22:24] <dolphy> ds-work: with no success
[22:24] <ds-work> yeah, if the pad is already connected, that won't work
[22:24] <ds-work> that will throw a g_return_if_fail() if you connect specific pads
[22:25] <ds-work> but if you just link elements, it automagically chooses pads that match, and if it can't match any, it gives up
[22:25] <ds-work> er, unconnected pads that match
[22:26] <dolphy> ds-work: yeah i understand that
[22:26] wheels ([email protected]) joined #gstreamer.
[22:26] <dolphy> ds-work: but does that sound possible that after unlinking autoplugger to video thread i m not able to link goom to video_thread because of some threading issue ?
[22:27] <dolphy> ds-work: what would be a solution here ?
[22:27] <ds-work> it would be nice if there was a simple tool to test out gst-player pipelines
[22:29] jowenn ([email protected]) joined #gstreamer.
[22:29] <dolphy> ds-work: that s what examples/gstplay is done for
[22:29] <jowenn> hi
[22:29] hadley ([email protected]) joined #gstreamer.
[22:31] <mathrick> hi hadley, jowenn, wheels; wb LeRoutier :)
[22:32] <wheels> hi mathrick 
[22:33] <dolphy> ds-work: i will try without ghosting
[22:33] <dolphy> ds-work: smells like a bug
[22:34] <markov> yoyo wheels 
[22:35] <apoc> eh shrek2 trailer plays fine with gst :)
[22:35] <mathrick> apoc: what format is it in?
[22:35] <Uraeus> apoc: mov ?
[22:36] <dolphy> ds-work: same issue... :-(
[22:37] <ds-work> dolphy: could be
[22:37] <ds-work> example/gstplay doesn't even work for me
[22:37] <ds-work> and it even screwed up rhythmbox
[22:37] <apoc> mathrick, Uraeus : mov (svq3 + mp3)
[22:37] <dolphy> ds-work: yeah i haven't commited yet
[22:37] <dolphy> ds-work: can that be because gst_element_unlink (autoplugger, video_thread) is not working
[22:37] <apoc> Uraeus: wma should work again now
[22:37] <mathrick> apoc: cool :)
[22:38] <dolphy> ds-work: something like autoplugger not ghosted correctly ..
[22:39] <dolphy> ds-work: maybe the pipeline need to be set to READY for that to work
[22:39] <Uraeus> apoc: so there is a small chance my drm wma file might work now :)
[22:40] <dolphy> ds-work: what's happening with examples/gstplay on your comp ?
[22:41] <apoc> Uraeus: I don't think so but you can try ;)
[22:43] <Company> Misirlou: pong
[22:43] <Company> ha, the shrek trailer is supposed to work, I fixed the A/V sync stuff with it :p
[22:44] <Company> (after only wanting to fix qtdemux but getting annoyed by the arrow hitting the tree way before it made that hitting sound)
[22:44] <Company> mathrick: you'Re not supposed to look inside an autoplugger and touch stuff in there
[22:45] <apoc> Company: :)
[22:45] <mathrick> Company: ok, so how is app using it supposed to tweak params of elems inside?
[22:45] <Company> mathrick: the way to get "preferences" for stuff the autoplugger does that I want to implement is changing ranks of plugins in a custom user registry
[22:46] <Company> mathrick: an app is not supposed to look inside the autoplugger - it either needs to plug itself or trust the autoplugger
[22:46] <mathrick> Company: huh? how does it map to elem properties?
[22:46] <Company> mathrick: not at all
[22:46] <mathrick> Company: i don't want to mess with what elems autoplugger has chosen, only to adjust some params of these elems
[22:47] <Company> mathrick: if you want to touch elements inside an autoplugger, write an autoplugger that allows this :)
[22:47] <Company> mathrick: i plan to add default values for properties to the registry that are user configurable
[22:48] <Company> mathrick: but I'm not going to support poking inside an autoplugger
[22:48] <mathrick> Company: hmm, default ones? May not be enough. let's say it's about hw iDCT, and some top-notch video engineer may decide hw is not accurate enough, so he'd like to choose sw one. How should it be exposed?
[22:49] <Company> mathrick: by an app that does this - a video engineer is not supposed to use an autoplugger
[22:49] <Misirlou> Company: The guy from bug 128098 gave me the MP3 in question.
[22:49] <Company> mathrick: you're not even guaranteed the autoplugger picks the elements you think it should pick
[22:50] <dolphy> Company: http://getcontrol.sourceforge.net/files/gstplay_pipeline.png
[22:50] <dolphy> Company: any idea why gst_element_unlink (spider, video_thread) is actually not unlinking them ?
[22:50] <mathrick> Company: hmm, AFAIK, things like Premiere do autoplug, so you have at least some pipeline to start with. Requiring user to plug everything manually is not what i'd call nice
[22:51] <Misirlou> Argh, now for some reason, I can't download it.
[22:51] <Company> dolphy: probably because noone ever did disconnecting and it therefore is buggy as hell ;)
[22:51] <Company> dolphy: but no, i have no clue
[22:52] <Company> mathrick: i bet when you develop a premiere competitor, you'll have the time to also write a pokable autoplugger
[22:52] <mathrick> Company: did old autoplugger work in a way that it had chosen elems, and after plugging, there was just a normal pipeline, with no autoplugger elem itself inside?
[22:53] <mathrick> Company: well, that's kind of true ;)
[22:53] <Company> mathrick: "old autoplugger"?
[22:55] <dolphy> Company: unlinking and linking spider might work ?
[22:55] <dolphy> Company: or do i need to put a new spider anyway ?
[22:56] <mathrick> Company: that one with autoplugcache
[22:56] <Company> dolphy: i wouldn't want to try it...
[22:56] <Company> mathrick: that one wasn't an element
[22:56] <dolphy> hmm i think i will simply throw visualization to trash :)
[22:57] <mathrick> Company: I know, it was mechanism, that's why I asked
[22:57] <Company> mathrick: you gave it sinkcaps and srccaps and it plugged elements for you
[22:57] <Company> mathrick: so afterwards you were left with a list of elements
[22:58] <mathrick> Company: AFAIK, that's how DS autoplugging works
[22:59] <mathrick> Company: this way is nice if you want more control than just 'play as if by magic'
[23:00] <Company> mathrick: that might be possible again with libgstreamer-autoplug in the future if you want to
[23:00] <mathrick> Company: btw, how is choosing elems on more than one element long path solved, and how do you plan to do it in new autoplugger?
[23:00] <ds-work> Company: what do you think about adding a "format-defined" int to GstData or GstBuffer?  It could be used for either subchannels (for float audio) or top/bottom interlaced images
[23:00] <Company> mathrick: but it's a stupid thing IMO
[23:00] <mathrick> Company: libgstreamer-autoplug <-- what's that?
[23:01] <dolphy> Company: ok so i confirm that unlinking autoplugger is broken
[23:01] <Company> mathrick: the new autoplugger that is currently in planning stages
[23:01] <dolphy> Company: getting the sink pad of the video thread and then its peer and unlink those pads work
[23:01] <Company> mathrick: "more than one element long path"?
[23:01] matrixise ([email protected]) joined #gstreamer.
[23:02] <Company> ds-work: there's already enough "format-defined" confusionn
[23:02] <mathrick> Company: like, when there is no element that transforms sinkcaps into srccaps directly
[23:02] <Company> mathrick: spider does shortest path algorithms
[23:03] <mathrick> Company: which one? Floyd-Warshall?
[23:03] <mathrick> Company: and how is graph of possible transformations built?
[23:04] <Company> mathrick: look at the code :)
[23:04] <mathrick> Company: heh, i hoped you'd spare me that ;)
[23:04] LeRoutier ([email protected]) left irc: Read error: 104 (Connection reset by peer)
[23:04] <Company> mathrick: i don't even know what "Floyd-Warshall" is ;)
[23:05] <Uraeus> apoc: does modplug also work again now?
[23:05] LeRoutier ([email protected]) joined #gstreamer.
[23:05] <mathrick> Company: the one computing shortest path from every to every node, using kind of relaxation scheme. I like to think of it as Dijkstra on steroids ;)
[23:06] <Company> mathrick: uh, i don't even remember how it works exactly - i only know the autoplugger is left with the shortest path and plugs it ;)
[23:06] <Company> mathrick: which doesn't work in real life, but that's another matter
[23:06] <mathrick> Company: heh :)
[23:06] <mathrick> Company: why?
[23:07] <Company> mathrick: wave files are plugged via wavparse
[23:07] <Company> so src ! wavparse ! audiosink
[23:07] <Company> but that breaks if there's an mp3 inside the wav
[23:07] <mathrick> Company: and how should they be plugged then?
[23:08] <Company> mathrick: select the element with "best chances" and plug that - do not select the element after that until you know the caps of the element you selected
[23:09] <Company> mathrick: it's useless to select anything in the wav case because wavs can contain anything
[23:09] <Company> mathrick: same for all other demuxers
[23:09] <mathrick> Company: wavparse is demuxer or decoder? I can't gst-inspect atm, installing gst-plugins right now
[23:09] <Company> mathrick: it _should_ be a demuxer
[23:09] <mathrick> Company: vorbisfile bis?
[23:10] <Company> a demuxer is an element that strips additional information and extracts streams
[23:11] <Company> the process of a demuxer is reversible in a lossless way
[23:11] <mathrick> Company: yeah, I know, I just meant if it does the same thing vorbisfile does, ie. demux and decode in one step
[23:12] <Company> it does not decode
[23:12] <mathrick> Company: it == wavparse? what does it do then that breaks for mp3?
[23:12] <Company> that's a misconception from the fact that data in wavs is normally not encoded at all
[23:12] <Company> mathrick: nothing
[23:13] <Company> mathrick: but shortest path assumes it outputs raw audio
[23:13] <Company> mathrick: when it outputs mp3
[23:13] <mathrick> Company: ahh, so it is demuxer, but advertises itself as being decoder too?
[23:13] <mathrick> Company: but that's really broken elem, not spider algo IMO
[23:14] <Company> mathrick: no
[23:14] <Company> mathrick: when spider looks at wavparse it looks at wavparses caps
[23:15] <Company> mathrick: at _all_ possible caps
[23:15] <mathrick> Company: ahh, I see it now
[23:16] <Company> shortest path does not take into account that data may limit the available caps on srcpads
[23:16] <mathrick> Company: hmm, "best chances" plugging sounds a bit tricky, not to say possibly slow
[23:17] <Company> why?
[23:18] <mathrick> Company: because it may lead into trying many elems before finding the one that does the trick. And when doing multistep transform, you don't really know what intermediate caps you should aim for
[23:19] <mathrick> Company: unless there's something more to it i'm not aware of
[23:20] <Company> mathrick: in your average decoding autoplugger, you get a caps and a limited set of elements (you don't use encoder, muxer or visualizer elements for decoding)
[23:20] <Company> mathrick: most of the time there's exactly one element that can be plugged (or two that do the same job)
[23:20] thaytoo ([email protected]) joined #gstreamer.
[23:21] <Company> mathrick: so there's no point in doing complex calculations, just take that one element
[23:21] <Uraeus> hi thaytoo
[23:21] <thaytoo> hi Uraeus 
[23:21] <Company> the logic is a bit more interesting if you want to do tagging ;)
[23:21] <mathrick> Company: hmm, right. But this one will fail miserably ASA this assumption isn't true, and Mr. Murphy is probably preparing such a case right now ;)
[23:22] <mathrick> yo thaytoo
[23:22] <mathrick> Company: what do you need to handle tagging?
[23:22] <Company> mathrick: that's why we code in a modular way - we'll just replace the element selection algorithm and try again :)
[23:22] <Company> mathrick: tagging an ogg file should autoplug oggdemux ! vorbistag ! oggmux
[23:22] <mathrick> Company: hehe :)
[23:23] <Company> mathrick: tagging an mp3 file should autoplug id3tag
[23:23] <Company> mathrick: tagging an id3 file should autoplug id3tag
[23:23] <mathrick> Company: why oggmux at the end?
[23:23] <Company> mathrick: tagging a flac file could plug flacdec ! flacenc :)
[23:23] <thaytoo> I'm sensing a pattern
[23:24] <Company> mathrick: vorbis needs a container
[23:24] <mathrick> Company: ah, ok, you mean tagging as in 'writing tags'
[23:24] <Company> mathrick: yeah
[23:24] <mathrick> Company: plugging encoding pipelines seems funky in general
[23:25] <Company> plugging encoding pipelines is a bit freaky
[23:25] <Company> because you can put anything into a file
[23:25] <mathrick> Company: yeah, that's what i mean
[23:25] <Company> and even if anyone says "avi" you can still put almost everything inside
[23:25] <dolphy> ds-work: filesrc location="/home/dolphy/matrix.avi" ! spider ! volume ! goom ! videoscale ! ximagesink
[23:25] <Company> which makes "auto"plugging nearly impossible
[23:26] <dolphy> ds-work: can you fix that one please ?
[23:26] <mathrick> Company: OTOH, is really encoding autoplugging that good idea?
[23:26] <dolphy> ds-work: adding removing videoscale make it working/non working
[23:26] <mathrick> Company: after all, we may end up with raw rgb inside avi
[23:27] <mathrick> Company: and noone really could say it's not 'proper' encoding pipeline
[23:27] <Company> mathrick: you want 3 autopluggers for video encoding
[23:27] <Company> mathrick: like videoin ! spider ! video/in-container ! spider ! file/format ! filesink
[23:28] <Company> same for audio
[23:28] <dolphy> ds-work: filesrc location="/home/dolphy/matrix.avi" ! spider ! volume ! goom ! ffcolorspace ! videoscale ! ximagesink
[23:28] <Company> so that the autoplugger only selects an element that does the raw video => encoded video and the muxing part
[23:28] <dolphy> ds-work: does exactly the same issue
[23:28] <mathrick> Company: hmm? i don't get that video/in-container part, and where's the 3rd spider?
[23:29] <Company> mathrick: video/in-container is the caps between encoder and muxer ;)
[23:29] <Company> mathrick: like video/x-xvid
[23:30] <mathrick> Company: ahh, so you meant video/out-container then ;)
[23:30] <Company> whatever :p
[23:31] Action: mathrick curses Digital Circuits
[23:31] Action: dolphy goes to bed
[23:31] <dolphy> night
[23:31] <mathrick> nite dolphy
[23:31] dolphy ([email protected]) left irc: "Network down, IP Packets delivered via UPS"
[23:31] <thaytoo> Uraeus: btw, http://masher.homeip.net/~jan/diary/ 
[23:32] Action: ds-work grrrs at dolphy
[23:33] <Uraeus> thaytoo: cool, ok just asked jeff to put it on planetgnome :)
[23:33] <thaytoo> hey
[23:33] <thaytoo> no!
[23:33] Action: ds-work reminds dolphy that apps *must* fixate caps or deal with the results
[23:34] <Uraeus> thaytoo: huh?
[23:34] <thaytoo> Uraeus: scary :)
[23:35] Action: Company suggests to ds-work that it might be a nice idea if videoscale would fixate to input size
[23:36] matrixise ([email protected]) left irc: "Leaving"
[23:37] <mathrick> thaytoo: now that you revealed URL, you're doomed even more than you'd be after putting your email on www.spamme.com ;)
[23:37] <thaytoo> yeah, it might have been a bit foolish
[23:37] <thaytoo> I may have to move the server ;)
[23:38] <mathrick> thaytoo: resistance is futile, they'll track you down ;)
[23:38] <ds-work> Company: actually, it's probably goom that should fixate
[23:43] <Uraeus> ok, time for bed, night
[23:43] Uraeus ([email protected]) left irc: "Client exiting"
[23:44] <LeRoutier> ++ got to kill myself
[23:44] <Company> ds-work: dunno, fixating is your invention :)
[23:44] LeRoutier ([email protected]) left #gstreamer ("Leaving").
[23:45] <ds-work> Company: it was a particularly crappy one at that
[23:45] <ds-work> Company: at least it keeps gst-launch from regressing into the stone age
[23:45] <ds-work> (mostly)
[23:47] mxpxpod ([email protected]) joined #gstreamer.
[23:49] Action: Jaramir is away: overflow error in /dev/null
[23:51] kmaraas ([email protected]) left irc: Read error: 110 (Connection timed out)
[00:00] --- Wed Jan 21 2004
[00:02] <thaytoo> nice one, rml: http://www.cafeshops.com/tarnished.9453818
[00:04] Action: ds-work cackles
[00:15] Nick change: harshyWork -> harshy
[00:16] <iain> oh I wish I still had an invitation to Nat's party where everyone had to bring an empty brown paper bag with "Illegal Drugs" written on it it crayon
[00:16] <iain> it was the bestest invitation ever
[00:17] <ds-work> bonus for actually _having_ illegal drugs?
[00:18] <iain> no, it had to be empty
[00:18] <iain> and the crayon had to have an interesting name
[00:18] <iain> like "Burnt umber"
[00:19] <iain> you weren't allowed a boring "red" for example
[00:19] <iain> and it had directions for getting there on foot, car, train, taxi, or helicopter
[00:20] <iain> and a list of words I use but am too embarassed to look up their actual meaning
[00:20] <iain> and a list of girls Nat sees frequently around boston but doesn't know their names, and who are timestoppingly gorgeous
[00:21] <iain> as well as emailing it to people, he printed it out ten times
[00:22] hadley ([email protected]) left irc: "Leaving"
[00:22] <iain> put it into eggshells, resealed the eggs and then swapped real eggs for the eggs with the party invitations in a supermarket at random
[00:22] <iain> sadly I missed the party :(
[00:23] <iain> thats my story finished...I'll ummm, go do something useful now
[00:26] <ds-work> taaz: ?
[00:27] <jowenn> Could it be that there is a problem using 2 libmad plugins in one pipe ?   2x filesrc ! spider ! passthrough -> adder -> osssink doesn't play anything with cvs for me and with gentoo's 0.6 version one stream stops after a few seconds
[00:27] <ds-work> potentially
[00:31] <ds-work> adder looks broken
[00:33] <ds-work> but multiple mads works
[00:36] kmaraas ([email protected]) joined #gstreamer.
[00:37] <jowenn> Is there any other working plugin to mix 2 or more streams into one ?
[00:39] <ds-work> switch
[00:39] <ds-work> it doesn't mix, it just switches
[00:45] <mathrick> ds-work: what about mixmatrix? It may be awfully complicated for the task, but at least may work
[00:45] <ds-work> mixmatrix hasn't been updated in forever
[00:46] <mathrick> that's what I suspected
[00:47] <jowenn> :(
[00:47] <jowenn> enough for today
[00:48] <jowenn> cu
[00:48] <mathrick> cya
[00:48] jowenn ([email protected]) left irc: "using sirc version 2.211+KSIRC/1.3.9"
[00:53] jowenn ([email protected]) joined #gstreamer.
[00:53] <jowenn> re
[00:54] <jowenn> With the old 0.6.x it is definitly the mad plugin, wav files work.
[00:54] <jowenn> ok, I'm really gone now
[00:54] jowenn ([email protected]) left irc: Remote closed the connection
[01:01] <ds-work> someone want to fix the line-splitting code in gst-inspect so it doesn't suck?
[01:03] <thomasvs> which one would that be ?
[01:04] <ds-work> the caps line splitting
[01:04] <ds-work> it absorbs the commas and looks ugly
[01:05] <ds-work> thomasvs: what do you think about Company's double-printf trick in email?
[01:07] markov ([email protected]) left irc: "nite"
[01:08] <thomasvs> ds-work: while it's a nice trick, it assumes/forces repeat of the message into the debug
[01:08] <ds-work> how so?
[01:09] <thomasvs> ds-work: let me reread it first to see if it's different from his first idea he pitched on irc
[01:09] <thomasvs> ds-work: any other objections on the error stuff btw ?
[01:09] <ds-work> well, i was thinking of it a bit differently
[01:10] <ds-work> basically, gst_element_error (..., "message %d", 1, "debug %d", 2)
[01:10] <thomasvs> ah, I see he fixed the thing I noted in his previous idea
[01:11] <thomasvs> hm, that looks nice, but it still is very different from what we're used to
[01:11] <thomasvs> ds-work: is it possible to make it work like you just said ?
[01:11] <ds-work> thomasvs: checking...
[01:11] <thomasvs> ds-work: also, what would be the way to pass a NULL using yours ? "" ?
[01:11] <ds-work> I *think* so
[01:11] <ds-work> NULL
[01:11] <thomasvs> you can't use NULL for a format
[01:11] <ds-work> you can if you check for it
[01:11] <thomasvs> hm - yeah, if it works for va_args
[01:12] Action: thomasvs wonders why he didn't try
[01:12] <thomasvs> I wouldn't mind at all if it was the way you say
[01:12] <thomasvs> ds-work: so if you send me the patch for the function and for a few errors in the core so I can test tomorrow, then I can do the rest of the conversion tomorrow during the day
[01:14] <ds-work> ok
[01:14] <ds-work> are you going to be up for a while?
[01:14] <thomasvs> no, that's why I propose you send me the small patch and I finish the big work :)
[01:15] <ds-work> ok
[01:15] <thomasvs> I really need some sleep, I'm feeling a bit sickish and our new hire is arriving tomorrow
[01:16] <thomasvs> who *are* all these people in here that never say anything ?
[01:16] <thaytoo> me!
[01:16] <thomasvs> no, we know you
[01:16] <thomasvs> we think we do
[01:16] <thomasvs> do we ?
[01:16] <mathrick> thomasvs: who is the new hire?
[01:16] <thomasvs> mathrick: some young nipper I know :)
[01:17] <mathrick> thomasvs: he doesn't hang around here?
[01:17] <ds-work> like pgoel, Rotty_, Shoragan, thread
[01:17] <mathrick> thomasvs: and getting some sleep isn't that bad idea ;)
[01:18] <thomasvs> mathrick: sometimes he does, right now he doesn't
[01:18] <mathrick> thomasvs: ic
[01:19] <mathrick> thomasvs: but after he arrives, take some time to show him around, ok? :)
[01:20] <mathrick> ok, nite to y'all, boys & girls
[01:20] <thomasvs> heh, that's the plan :)
[01:20] Nick change: mathrick -> mathrick|sleep
[01:20] <thaytoo> they're interested people just WAITING for their chance to hack code for gstreamer!
[01:20] <thaytoo> and some are here to write documentation
[01:21] yippi ([email protected]) left irc: "Client exiting"
[01:26] <ds-work> thomasvs: hrm, my idea doesn't work
[01:27] <ds-work> unless we want to write our own printf function
[01:28] <thomasvs> why not ? :)
[01:29] <ds-work> because vprintf() takes va_list and not va_list *
[01:29] <ds-work> Company's idea will work, though
[01:30] <ds-work> i.e., gst_element_error (..., "message %d", "debug %d", 1, 2)
[01:30] Nick change: kh_zZz -> khalek
[01:30] <ds-work> except that you may have to live with the debug message being "message 1 debug 2"
[01:31] <ds-work> hrm
[01:31] <ds-work> although that could be split at strlen(message)
[01:34] <thomasvs> ds-work: that's the thing - I don't see why we should signal the message twice
[01:35] <thomasvs> but I don't see how (format1, args1), (format2, args2) is any worse/less standard then format1, format2, args1, args2
[01:36] maihde ([email protected]) joined #gstreamer.
[01:37] maihde ([email protected]) left irc: Client Quit
[01:37] <ds-work> thomasvs: because I want things that look like functions to act like functions
[01:50] <ds-work> thomasvs: http://www.schleef.org/~ds/double_printf.c
[02:01] alley_cat ([email protected]) left irc: "May the Source be with you!"
[02:08] sub_pop ([email protected]) joined #gstreamer.
[02:16] foser ([email protected]) left irc: "[ I want to believe ]"
[02:22] <Company> i like the fact that the debug message includes the error mesaage
[02:23] <Company> the only problem is that the debug message is translated...
[02:23] <Company> s/debug message/error part of the debug message/
[02:24] <ds-work> Company: it doesn't have to -- the code that I posted automatically splits the strings
[02:25] <Company> i know
[02:25] <thomasvs> the debug message isn't translated, the error message is
[02:25] <thomasvs> but I still see no point at all in having the error message in the debug as well
[02:25] <Company> because you might want to GST_DEBUG the debug message
[02:25] <thomasvs> if the app wants to do that, fine
[02:26] <thomasvs> all errors get GST_DEBUG'd by default anyway
[02:26] Action: Company still fears debug messages like this: "17"
[02:26] <thomasvs> why would you have a debug message like that ?
[02:26] <Company> because someone thought the error message described the rest...
[02:26] <Company> Avidemux: unknown audio format
[02:26] <thomasvs> then that's a bug and it needs to be fixed
[02:27] <thomasvs> it's not really a valid point in favour of duplication of information :)
[02:27] <Company> i don't see it as duplication
[02:27] <Company> because it targets different audiences
[02:27] <thomasvs> there's no way to see A, A + B as not being duplication
[02:28] <thomasvs> anyways.  if you have a use case where it's actually useful that doesn't involve everyone pretending not to know anything, I'm glad to hear it
[02:28] <Company> well, if you force everyone to look at the error message then yes, it's duplication
[02:28] <thomasvs> otherwise, I see no point in doing A, A + B
[02:28] <thomasvs> Company: no one is going to see the debug message without having first seen the error message
[02:28] <thomasvs> anyways, night
[02:28] <Company> i just don'T see the error message as something i should look at
[02:28] Nick change: thomasvs -> thomasvz
[02:28] <thomasvz> ok, then don't look at it
[02:28] <Company> do you GST_DEBUG the error message, too?
[02:28] <thomasvz> then you don't want it repeated in the debug either :)
[02:28] <thomasvz> yes
[02:29] <Company> the translated one?
[02:29] <thomasvz> afaik yes
[02:29] <Company> hmhm
[02:30] <Company> should be ok though, the debug message will have every important information anyway
[02:31] <Company> because decs will just stick it in there if they think they need it
[02:31] Shoragan ([email protected]) left irc: "Leaving"
[02:31] <Company> devs
[02:32] <Company> and if we did the double_printf trick (or just use the same arguments twice, i don't care which we use though i'd prefer using the arguments twice) should be done
[02:32] <Company> and then we'Re all happy i guess :)
[02:35] <ds-work> I don't think format1,args1,format2,args2 can be done without a special vprintf function
[02:35] <ds-work> format1,format2,args1,args2 works
[02:35] <ds-work> is that what you mean?
[02:35] <ds-work> and then gst_element_error() can be a function again
[02:36] <Company> yeah, that's what i mean
[02:37] <Company> and I prefer a macro because i want __LINE__ and __FILE__ in there :)
[02:37] <ds-work> oh yeah
[02:37] <ds-work> no big deal there
[02:37] <ds-work> "macro that acts like a function" and "function" are the same in my mind
[02:38] <Company> and a GST_ERROR_OBJECT should be done automagically
[02:38] <Company> obviously in the default category
[02:41] <ds-work> hrm, rhythmbox doesn't come up with some translations
[02:42] <Company> have we solved the how-do-plugins-initialize i18n yet?
[02:46] iain ([email protected]) left irc: "ping png"
[02:46] <ds-work> grrr
[02:46] <ds-work> how do I get translations for apps installed in /usr/local/?
[02:47] <Company> dunno
[02:47] <Company> don'T they go into datadir/locale by default?
[02:48] <ds-work> yes
[02:48] <ds-work> but it's not getting picked up
[02:50] Action: Company notes that the vaporware awards do not include anything open source
[02:51] <Company> did you look at my autoplug stuff yet?
[02:51] <ds-work> no
[02:52] <Company> it's just headers and no code anyway
[02:52] <Company> i'd just like to know what you think about it
[02:52] <ds-work> in case you hadn't noticed, I've called subclasses of GstPlug after characters from finding nemo
[02:53] <ds-work> er, but I may have checked it in to the wrong place :)
[02:53] <Company> and she's called Dorie, not Dory :p
[02:54] <ds-work> not according to the box
[02:54] <ds-work> and imdb
[02:54] <ds-work> uh, where did your code go?
[02:54] <Company> then imdb changed it
[02:55] <Company> correct cvs?
[02:56] <Company> wrong cvs?
[02:56] <ds-work> I think wrong
[02:56] <ds-work> i.e., sf.net
[02:56] <Company> http://article.gmane.org/gmane.comp.video.gstreamer.cvs/11807
[02:57] <ds-work> ok, nm
[02:57] <ds-work> my checkout was in some very strange state


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn