[gst-devel] Daily IRC logs

wim.taymans at chello.be wim.taymans at chello.be
Wed May 2 06:27:55 CEST 2001

[08:17] dichro (dichro at foo.optusnet.com.au) joined #gstreamer.
[08:17] <omega_> yo
[08:17] <dichro> morning all
[08:17] Action: omega_ is writing up a painfully detailed explaination of INCSCHED1
[08:18] <dichro> and if you were asked to describe it as a single broad sweeping generalization, what would it be?
[08:18] <omega_> incremental scheduling <g>
[08:19] <dichro> ah :) as opposed to...?
[08:19] <omega_> um, "the ability to optimially maintain a viable schedule while modifying the pipeline while it's running or otherwise not completely shut down"
[08:20] <omega_> HEAD only generates a schedule from NULL->READY, which kinda makes things like autoplug rather nasty
[08:20] <dichro> where "schedule" is...? the whole pulling buffers around between cothreads thing?
[08:20] <omega_> mpeg2parse3 in INCSCHED adds audio and video parts of the pipeline without even PAUSEing anything
[08:21] <omega_> yeah, the schedule consists of basically two things: the wrapper that is the outer cothread context, and the function pointer proxies that implement gst_pad_pull and gst_pad_push
[08:21] Action: dichro nods
[08:21] <omega_> and eventually it'll be smarter again, and use chain'd elements on the stack, like they're supposed to be
[08:21] Action: dichro received a very nasty surprise when he discovered that gstreamer used pthreads liberally.
[08:21] <omega_> liberally?
[08:21] Action: dichro shrugs
[08:21] <omega_> what kind of surprise?
[08:21] <dichro> liberally enough to make my code crash nastily with very surprising bugs :)
[08:22] Nick change: aj_busy -> ajmitch
[08:22] <omega_> ah
[08:22] <dichro> then I realized that gdb was muttering something about lwp 1024, and it all fell into place :)
[08:22] <omega_> yeah, well, that's the other half of stuff that's landed in the INCSCHED1 branch: significantly upgraded thread locking semantics
[08:22] Action: dichro nods
[08:22] <omega_> matth and I have been working a lot on those issues
[08:23] <omega_> for instance, state changes to THREADs now are properly interlocked
[08:23] <omega_> we're working on mechanisms that keep deadlock from happening as well, when you try to switch to PAUSED state across thread boundaries
[08:23] <dichro> between that, v4lsrc freezing my box hard, and beating on the jpegenc plugin until it did something useful, it's been an interesting experience :)
[08:23] <omega_> you have patches?
[08:24] <dichro> I haven't split it out as patches - there's still teensy problems. Have a look at http://mikolaj.tv/ and see if you can spot one :)
[08:24] <dichro> (only got it working last night - I'll do up some patches tonight, hopefully, and submit... to whom? mailing list?)
[08:24] <omega_> yeah, -devel
[08:24] <omega_> what kinds of changes did you have to make?
[08:24] <omega_> eesh
[08:24] <omega_> yo
[08:25] <omega_> which one is more live, it seems that the b&w is
[08:25] <dichro> took a ruddy great big chainsaw to gstjpegenc.
[08:25] Action: dichro nods
[08:25] <omega_> this is done with gstreamer?
[08:25] <dichro> the b&w is running out of camserv, and sending ~7k frames
[08:25] <dichro> the colour one is gstreamer, and sending ~50k frames (I think)
[08:25] <omega_> yikes
[08:26] <omega_> what's up with the colors??
[08:26] <omega_> U and V swapped somehow?
[08:26] <dichro> still need to do a fair bit of work on gstjpegenc to allow it to specify compression params
[08:26] Action: dichro grins
[08:26] <omega_> or are you really blue?
[08:26] <dichro> and that's the problem :)
[08:26] <dichro> I think R and B are swapped.
[08:26] <omega_> aren't you passing yuv to jpegenc?
[08:26] <omega_> or doesn't it like taking that format?
[08:27] <dichro> it does, but I have no idea what it's supposed to be. I more or less think I understand RGB :)
[08:27] <dichro> I had to force jpegenc to demand a format, otherwise it wouldn't negotiate with v4lsrc, so I chose that at random
[08:28] <dichro> now I've gone and read a little about the ov511 driver (which has been going for *minutes* without crashing my box! yay!) and it natively grabs as YUV420, so I think it probably makes sense to go back and change everything to use that.
[08:28] <dichro> libjpeg groks a number of formats, including that, I think. Is one better than the other in any way?
[08:29] <omega_> yuv420 is preferable, since that's as close as it gets to native format
[08:29] <omega_> but jpeg uses non-standard yuv colorspace, so there's still a conversion
[08:29] <omega_> and I dunno whether it actually does 4:4:4 or 4:2:0, I should find out
[08:30] <omega_> and you hav ov511 working without crashing?
[08:31] <omega_> that took some changes.... it wedges with v4lsrc on my box
[08:31] <dichro> I grabbed 1.38 from the website - it's been doing a hell of a lot better than the stock 2.4.4 one :)
[08:31] <dichro> 1.37 is listed as fixing a number of crashes
[08:31] <omega_> hmm, ok
[08:31] Action: omega_ needs to get himself one of those things
[08:31] <omega_> which do you have?
[08:31] <dichro> I think there were about four separate ways of taking the box down via gstreamer with 1.34
[08:32] <dichro> the camera?
[08:32] <omega_> doh
[08:32] <omega_> yeah
[08:32] Action: omega_ tried a D-link at work
[08:32] <dichro> it's a D-Link DCS-C300 or something. hang on.
[08:32] <dichro> ov511.c: camera: D-Link DSB-C300
[08:32] <omega_> cool, I should get one then
[08:32] <dichro> there's a cheaper C100, but it wasn't supported by the driver at the time that I was playing with one. Apparently they are now, though.
[08:33] <dichro> they're quite cheap - sub A$100 here, so probably ~US$40 or so.
[08:33] <omega_> checking
[08:33] <omega_> wow, $41
[08:34] <ajmitch> wow, an aussie ;)
[08:34] <ajmitch> hi
[08:34] <dichro> One of my back burner projects is hooking up several to a single box and seeing how things go.
[08:34] <omega_> some of your gallery shots made me think that the colors were messed up, then I realized that it was just someone's hair.....
[08:34] <omega_> I should try that at work
[08:34] <dichro> ajmitch: hi :) are you likewise antipodal?
[08:34] <ajmitch> New Zealand
[08:34] <dichro> near enough, I guess.
[08:35] <ajmitch> ;)
[08:35] <dichro> omega: that would be Duncan, yes :)
[08:35] Action: omega_ wants a D300, it does 640x480
[08:36] <dichro> I think that's what this one does.
[08:36] <dichro> hang on, I'll try to crank up the resolution on egocam...
[08:37] <omega_> is that the gstreamer version?
[08:37] <dichro> yep
[08:38] <omega_> erm, did it crash?
[08:39] <omega_> doh, we lost him
[08:39] Action: omega_ watched until the end, saw the flash on his screen and the pained look
[08:40] dichro (dichro at foo.optusnet.com.au) left irc: Ping timeout for dichro[foo.optusnet.com.au]
[08:45] omega_ (omega at omegacs.net) left irc: Read error to omega_[omegacs.net]: EOF from client
[08:46] omega_ (omega at omegacs.net) joined #gstreamer.
[08:46] <omega_> um
[08:46] <omega_> my X server simply restarted
[08:46] <ajmitch> bad...
[08:47] <omega_> quite
[08:51] dichro (dichro at foo.optusnet.com.au) joined #gstreamer.
[08:51] <omega_> welcome back...
[08:52] <dichro> hmm. ov511 isn't totally stable yet :)
[08:52] <ajmitch> hehe
[08:52] <omega_> I watched your screen flash, the pained look on your face, and then tried in vain to ping you
[08:52] <ajmitch> i guess something went wrong? ;)
[08:52] Action: omega_ suggests a log-structured filesystem to dichro
[08:53] <dichro> yep. The same thing has occurred to me any number of times :)
[08:53] Action: omega_ uses reiserfs, hasn't seen longer than a 14sec fsck
[08:53] <omega_> for a 18GB filesystem
[08:53] <dichro> I keep meaning to try reiserfs, but I have a taste for non-intel architectures, and I'm a bit wary of reiserfs for that reason.
[08:53] <omega_> hmm
[08:53] <dichro> this is a pc, but it's a work machine
[08:54] <omega_> are you doing this project for work, or just screwing around with it? <g>
[08:54] <ajmitch> what non-intels archs? ppc, or others?
[08:54] <dichro> omega: this is potentially future business (although not for work), but right now mostly for pleasure
[08:54] <omega_> cool
[08:54] <dichro> ajmitch: sparc and alpha, for the most part
[08:54] <dichro> haven't played with ppc yet
[08:54] <omega_> linux or native os?
[08:54] <dichro> linux, naturlich :)
[08:54] <omega_> heh
[08:54] <ajmitch> ;)
[08:55] <ajmitch> what else? ;)
[08:55] <omega_> have you tried gstreamer on sparclinux yet?
[08:55] <dichro> nope - it bombs horribly on the alpha, though
[08:55] <omega_> oh?
[08:55] <ajmitch> dichro: where in australia are you?
[08:55] <dichro> ajmitch: Sydney
[08:55] <ajmitch> ahhh....
[08:55] <dichro> omega: now that I think of it, I can't quite remember where.
[08:56] <dichro> I spent a short while looking at it, but that was about three or four weeks ago
[08:56] <omega_> ok
[08:56] <dichro> I'll have another look at home when I get the chance
[08:56] <omega_> I successfully got a it compiled on my UDB, but didn't test it extensively
[08:56] <dichro> it may well have been me using v4lsrc and jpegenc in original forms - I really don't remember :(
[08:56] <omega_> hehehe
[08:57] <dichro> um. sink philosophy/
[08:57] <omega_> my testing pretty much ended after ./gstreamer-launch fakesrc ! fakesink worked
[08:57] <dichro> egocam is basically just the network handling code.
[08:57] <omega_> yup
[08:57] <dichro> do you think it makes sense for a plugin to encapsulate all the multi-client HTTP code inside it?
[08:57] <omega_> hrm, I dunno
[08:58] <dichro> I didn't think so originally, but the more I play, the more it appeals.
[08:58] <omega_> it could be, yeah
[08:58] <dichro> can I assume event handling inside a plugin?
[08:58] <omega_> if you're in your own thread, you can take it over
[08:58] <dichro> ie, can I assume that I'll always be running inside a gdk event loop?
[08:58] <dichro> ah - that would make more sense, yes :)
[08:58] <omega_> nope, if you're in a thread, you run in a private loop
[08:58] <omega_> but...
[08:58] <omega_> if it would make sense to make the thread use a glib main loop, we can certainly look into that
[08:59] Action: dichro nods
[08:59] <omega_> wouldn't be very hard, I don't think
[08:59] <dichro> the other thing is that this method isn't terribly generic - multipart/x-mixed-replace only works with netscape, not IE.
[08:59] <omega_> but it does have potential consequences for the cothreaded model
[08:59] <omega_> IE doesn't do multipart????
[08:59] Action: omega_ puts another hole through his Windoze CD
[09:00] <dichro> I'm not religious about Gdk events - they make things easier, but I've written event driven state machines far too often in the past to be fussed :P
[09:00] <omega_> heh
[09:00] <omega_> I will think about how they can be merged though, since I can see the benefits
[09:01] <dichro> I'll post some patches to the list for consideration - not any of this stuff, but the jpegenc changes and fakesink tweak as they stand.
[09:01] <omega_> ok
[09:01] <dichro> enough to duplicate what's running right now, as crippled as it is :)
[09:01] <omega_> hehe
[09:02] <dichro> one interesting thing....
[09:02] <omega_> ?
[09:02] <dichro> I'm definitely getting more data sucked out through gstreamer than I was through camserv.
[09:02] <omega_> hrm
[09:02] <omega_> efficiency?
[09:02] <dichro> ie, running encoder and browser on the same machine, it's pulling more data (in terms of k/sec) now than it was with camserv
[09:03] <omega_> neat
[09:03] <dichro> could be the large image size - although I wouldn't expect that either encoding or decoding jpeg images would scale sub-linearly
[09:03] <omega_> right
[09:06] dichro (dichro at foo.optusnet.com.au) left irc: Ping timeout for dichro[foo.optusnet.com.au]
[09:06] <omega_> oops, he killed it again
[09:08] <ajmitch> omega_: you have webcam?
[09:08] dichro (dichro at foo.optusnet.com.au) joined #gstreamer.
[09:09] <dichro> back to camserv :)
[09:09] <omega_> heh
[09:09] <dichro> for some reason, camserv *never* crashes my box.
[09:09] <omega_> odd
[09:09] Action: dichro looks around for some wood to knock on in a panic
[09:09] <ajmitch> haha
[09:09] <omega_> hrm, you look different all of the sudden....
[09:10] <dichro> all flushed and rosy-cheeked, you mean? :)
[09:10] <omega_> yeah, less sickly
[09:10] <ajmitch> probably some of that cheap aussie beer ;)
[09:10] <omega_> now we have to find a way for you to not swap places with the water cooler with egocam
[09:11] <dichro> I tried to get R and B swapped, but v4l was having none of it. Either it didn't accept the negotiation, or the v4l driver refused to accept the parameters
[09:12] <omega_> v4l driver can't do that
[09:12] <dichro> so something's getting shuffled somewhere else.
[09:13] <dichro> I had a heap of debugging code sprinkled throughout which dumped the frame at various stages of processing. I should turn it back on and see where the colour gets confused.
[09:13] <omega_> yeah
[09:13] Action: omega_ looks through the jpeg standard to see how effectively it can take YUV input
[09:14] <omega_> oh wait.
[09:14] <omega_> duh
[09:14] <omega_> jpeg is rgb by default, since it's not 'video'
[09:16] <omega_> but according to gstjpegenc.c, ycrcb jpeg is possible,the question is whether anyone like netscape can decode that
[09:20] steveb (chatzilla at 212186169160.chello.com) joined #gstreamer.
[09:20] <omega_> yo
[09:22] <dichro> there doesn't seem to be anything in the libjpeg docs about formats other than RGB
[09:22] <steveb> yo
[09:22] <ajmitch> hey steveb
[09:23] <steveb> omega_: so, what do you have against XSLT? :)
[09:23] <omega_> Special color spaces
[09:23] <omega_> steveb: nothing, it just sounds a little scary
[09:23] <steveb> so does megamacros
[09:24] <omega_> true
[09:24] <omega_> but I'm a macro fiend, don't you get it?
[09:24] <steveb> heh
[09:24] <omega_> just look at libcodec....
[09:25] Action: ajmitch doesn't want to lose his grip on sanity...
[09:25] <ajmitch> ;)
[09:25] <omega_> dichro: libjpeg 6b will do conversions to/from YCrCb and RGB
[09:26] <steveb> this guy has done all of his ladspa plugin src as xml + preprocessor - looks impressive
[09:26] <steveb> http://plugin.org.uk/
[09:26] <omega_> which means that as long as we can get the input format to match what libjpeg expects, we can go straight from v4l yuv to jpegenc
[09:26] <omega_> but the question remains: can browsers handle these kinds of jpegs?
[09:26] <dichro> omega: just reading that section now
[09:26] <dichro> but doesn't that mean that libjpeg just converts to RGB internally?
[09:26] <omega_> nope
[09:27] <omega_> jpeg stores arbitrary color channels of any size you want
[09:27] <omega_> but it's saying that it will convert yuv420 to rgb if you open a yuv420 jpeg and ask for it as rgb
[09:27] <dichro> right, okay.
[09:28] <omega_> that's what the x_samp_factor stuff is for, to describe the 420 stuff
[09:28] <omega_> which means that it's theoretically capable of arbitrary yuv formats
[09:29] <dichro> hmm. this actually seems quite reasonably. If I'm reading it right, I just have to say 'in_color_space = JCS_YCbCr' and it should be fine.
[09:29] <dichro> reasonable, that being.
[09:30] <omega_> right, as long as the actual in-core format jpeg expects is what you're giving it, which is very iffy
[09:31] <dichro> hrm. I get the feeling that it still expects 3 8-bit samples per pixel...
[09:31] <dichro> What *is* YUV? :)
[09:31] <omega_> not in ycrcb mode, but I doubt it uses the same interface as rgb
[09:31] <omega_> yuv is luminance, chrominance
[09:31] <omega_> Y is luma, or B&W channel
[09:31] <omega_> U and V (or Cr, Cb) are differences that embody the red and blue channels
[09:32] <dichro> right. thanks :)
[09:32] <omega_> green isn't coded explicitely because it's derived from Y and the differences, since green provides most of the luminance data
[09:32] <omega_> the key with yuv is that U and V are usually spatially subsampled
[09:33] <omega_> so 4:2:0 has a luma sample per pixel, but one Cr and Cb sample for every 2x2 block of pixels
[09:33] <omega_> and of course you can arrange things either packed or planar
[09:33] <omega_> see www.webarts.com/fourcc/ for a list of known YUV formats
[09:35] <dichro> ah. okay, that would blow the fixed sample size idea out of the water.
[09:35] <omega_> right
[09:37] <omega_> it appears that gstjpegenc as it lives in CVS is designed to only handle planar 4:2:0
[09:37] <omega_> and not very efficiently either
[09:37] <dichro> planar meaning that you pad out the samples to be fixed size?
[09:38] <omega_> no, planar meaning that the entire Y channel is linear
[09:38] <omega_> then the Cr and Cb channel images are concatenated
[09:38] <omega_> for image of 640x480, you have 640*480 Y samples, then 320*240 Cr, then 320*240 Cb samples
[09:38] <dichro> erm - are we talking about the bit ordering in a sample?
[09:39] <dichro> ah, right.
[09:39] <omega_> nope, sample ordering it the image
[09:39] <omega_> packed would be YYcrYYcb or something
[09:39] Action: dichro finally gets the idea :)
[09:40] <omega_> afaict there are no packed 4:2:0 formats anyway, so it's kinda irrelevant
[09:41] <dichro> *blink*
[09:41] <dichro>  * Select an appropriate JPEG colorspace for in_color_space.
[09:41] <dichro>   case JCS_RGB:
[09:41] <dichro>     jpeg_set_colorspace(cinfo, JCS_YCbCr);
[09:42] <omega_> one problem is that YUV != YCrCb, necessarily
[09:42] <omega_> http://www.inforamp.net/~poynton/notes/colour_and_gamma/ColorFAQ.html
[09:43] <dichro>     /* We default to 2x2 subsamples of chrominance */
[09:43] <dichro> does that sound like 4:2:0?
[09:43] <omega_> yup
[09:44] <dichro> okay, then that would be the code to set up arbitrary subsample sizes.
[09:44] Action: dichro wonders if that's for input or output.
[09:45] <omega_> I would assume it depends on which struct you fill in
[09:45] <omega_> or whether you filled it in yourself or not <g>
[09:45] <dichro> it's definitely fiddling with input_components in some places. There is hope :)
[09:46] <dichro> hmm. so, theoretically, if you received a streamed planar YCbCr image, you could render it in grayscale first and then fill in the colour?
[09:46] <omega_> sorta, yeah
[09:47] <dichro> interesting
[09:47] <dichro>       cinfo->jpeg_color_space = JCS_YCbCr; /* JFIF implies YCbCr */
[09:47] <dichro> all of the jpgs that I'm emitting at the moment are tagged as JFIF.
[09:47] <omega_> afaict, YUV was invented during the transition to color TV
[09:48] <omega_> JFIF is the file format, nothing to do with the color space
[09:48] <omega_> that comment makes no sense
[09:48] <dichro> relevant bits:
[09:48] <dichro>  * Set default decompression parameters.
[09:48] <dichro> default_decompress_parms (j_decompress_ptr cinfo)
[09:48] <dichro>   /* Guess the input colorspace, and set output colorspace accordingly. */
[09:48] <dichro>   /* (Wish JPEG committee had provided a real way to specify this...) */
[09:48] <dichro>   /* Note application may override our guesses. */
[09:48] <dichro>   switch (cinfo->num_components) {
[09:48] <dichro>   case 3:
[09:48] <dichro>     if (cinfo->saw_JFIF_marker) {
[09:48] <dichro>       cinfo->jpeg_color_space = JCS_YCbCr; /* JFIF implies YCbCr */
[09:48] <dichro>     } else if (cinfo->saw_Adobe_marker) {
[09:48] <dichro>       switch (cinfo->Adobe_transform) {
[09:49] <dichro>       case 1:
[09:49] <dichro>         cinfo->jpeg_color_space = JCS_YCbCr;
[09:49] <dichro>     } else {
[09:49] <dichro>       /* Saw no special markers, try to guess from the component IDs */
[09:49] <omega_> hrm, it could mean that JFIF implies ycbcr if there's no other APPn marker to explicitely say which space it's actually in
[09:50] <dichro> I think the JFIF is described as being an APPn marker
[09:50] <dichro>   if (cinfo->write_JFIF_header) /* next an optional JFIF APP0 */
[09:50] <dichro>   if (cinfo->write_Adobe_marker) /* next an optional Adobe APP14 */
[09:50] <omega_> hrm, everything I've read says that JFIF is the outer file format, and that JPEG actually doesn't specify an actual file format
[09:51] <omega_> http://www.faqs.org/faqs/jpeg-faq/part1/
[09:51] <omega_> subject 15
[09:53] <dichro> hmm.
[09:53] <dichro> I am far from qualified to comment :)
[09:54] <omega_> I'm tempted to write a new jpeg lib <g>
[09:55] <dichro> No, damnit! Make gstreamer all singing and all dancing first! :)
[09:55] Action: omega_ hands gstreamer a beer
[09:55] <omega_> that should help some....
[10:06] Action: omega_ will be back in ~15min, then is going to sleep
[10:06] <dichro> google has a number of negative things to say about libjpeg and YUV
[10:06] <omega_> oh?
[10:07] <omega_> what query?
[10:07] <dichro> v4l's YUV apparently being incompatible with libjpeg's as you suggested.
[10:07] <dichro> libjpeg yuv :)
[10:07] <omega_> depends on the fourcc, mostly
[10:08] <omega_> well, I'll bbr
[10:39] Action: omega_ needs to sleep
[10:39] omega_ (omega at omegacs.net) left irc: killall -9 omega
[10:51] Ow3n (owen at ti34a80-0313.bb.online.no) joined #gstreamer.
[10:52] steveb (chatzilla at 212186169160.chello.com) left irc: ChatZilla 0.8 [Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; 0.8.1) Gecko/20010323]
[10:56] thomas (thomas at 212-100-172-175.adsl.easynet.be) joined #gstreamer.
[11:04] dichro (dichro at foo.optusnet.com.au) left irc: Leaving
[11:20] ajmitch (ajmitch at p58-max11.dun.ihug.co.nz) left irc: Ping timeout for ajmitch[p58-max11.dun.ihug.co.nz]
[11:28] ajmitch (ajmitch at p58-max11.dun.ihug.co.nz) joined #gstreamer.
[11:52] ajmitch (ajmitch at p58-max11.dun.ihug.co.nz) left irc: Ping timeout for ajmitch[p58-max11.dun.ihug.co.nz]
[11:58] ajmitch (ajmitch at p58-max11.dun.ihug.co.nz) joined #gstreamer.
[12:55] Nick change: maYam_sleep -> maYam
[13:18] richardb (richard at ixion.tartarus.org) joined #gstreamer.
[13:19] <thomas> richardb: you're working on the floating point audio, right ?
[13:31] Nick change: wtay-zZz -> wtay
[13:31] <wtay> yo
[13:32] <maYam> morning, wtay
[13:32] <wtay> doh
[13:33] <maYam> thanks for the coffee
[13:34] <wtay> maYam: no prob :)
[13:38] <thomas> you two aren't sitting next to each other right ;) ?
[13:38] wtay (wim at cable-195-162-214-190.upc.chello.be) left #gstreamer (X restart).
[13:38] <maYam> yes we are ;)
[13:39] <thomas> heh
[13:44] <richardb> thomas: documenting it, yep.
[13:45] <richardb> Hmmm, I must teach louise to use IRC...
[13:45] <maYam> no richardb.. real communication has its charms as well ;)
[13:47] <richardb> true
[13:48] thomas (thomas at 212-100-172-175.adsl.easynet.be) left irc: Read error to thomas[212-100-172-175.adsl.easynet.be]: EOF from client
[13:52] wtay (wim at cable-195-162-214-190.upc.chello.be) joined #gstreamer.
[14:20] thomas (thomas at 212-100-172-175.adsl.easynet.be) joined #gstreamer.
[14:20] <thomas> hi
[14:24] Gandalf_ (gandalf at tux.rsn.bth.se) left irc: 
[14:25] <wtay> yo thomas
[14:27] <richardb> Anyone know what valid values for the layout property are
[14:27] <wtay> richardb: for float?
[14:27] <richardb> (for audio/raw, format=float)
[14:27] <richardb> I'm presuming it's only relevant for format=float
[14:27] <richardb> gfloat appears to be one
[14:27] Action: thomas tries to fix vmware's choppy audio
[14:29] <thomas> wtay: I read the logs... is it possible to output the pipeline structure of the mixer and see it in the editor ?
[14:30] <wtay> thomas: I had good results, YMMV
[14:31] <wtay> richardb: gfloat is an option I don't really know how much that describes the format though..
[14:31] <thomas> wtay: what's YMMV ?
[14:32] <wtay> Your Millage May Vary
[14:32] <thomas> ok, can I copy the necessary code from an example and try it out ? because I'm not sure about the mixer setup anymore
[14:33] <wtay> thomas: just perform gst_xml_write () on the toplevel bin
[14:33] <wtay>   xmlSaveFile("xmlTest.gst", gst_xml_write(GST_ELEMENT(bin)));
[14:43] <thomas> wtay: hmm, ok, that seems to work.
[14:44] <thomas> first observation : my adderel and play_audio are in a bin unconnected to the rest !
[14:44] <thomas> doesn't look good does it
[14:45] <wtay> hmm
[14:45] <wtay> depends, it it never gets connected at runtime, it's bad
[14:47] <thomas> wtay: I did the xmlSaveFile after the first gst_pad_connect, so I guess that it is unable to connect.
[14:47] <thomas> Could that be the case ?
[14:47] <wtay> yes
[14:52] <thomas> wtay: can the editor show ghostpads ?
[14:52] <wtay> thomas: nope
[14:52] <thomas> wtay: ok so there's no hope of having the editor display the real situation then ?
[14:52] <thomas> because the info output for mixer shows a good connect :
[14:52] <wtay> thomas: hmm, yeah, true
[14:52] <thomas> INFO ( 2383: 0):gst_pad_connect:636: connected pipeline1:channel1 and adderel:sink0
[14:53] <thomas> so that looks good
[14:54] dichro (dichro at foo.optusnet.com.au) joined #gstreamer.
[14:56] <thomas> wtay: so the xml generated with xmlsave doesn't take ghostpads into account either ?
[14:56] <wtay> thomas: hmmm, good question. I doubt it does actually...
[14:57] <thomas> wtay: ok, there must be a way to debug the connect problem.
[14:57] <thomas> i'm going to rewrite it from the bottom
[14:57] <thomas> luckily the top is close away ;)
[14:57] <wtay> thomas: no indication of nego failed?
[14:58] <thomas> wtay: how can I see that in the output ? it's rather large.
[14:58] <thomas> can I search on some string ?
[14:59] <wtay> thomas: failed?
[14:59] <thomas> no failed anywhere
[15:00] <wtay> good and bad at the same time <g>
[15:01] dichro (dichro at foo.optusnet.com.au) left irc: Leaving
[15:08] <thomas> wtay: can you set info or debug mask within code ?
[15:08] <thomas> so that I can skip all the messages I don't need anyway
[15:08] <wtay> thomas: yup, sec..
[15:08] <wtay> gst_info_set_categories         (guint32 categories);
[15:09] <wtay> gst_debug_set_categories        (guint32 categories);
[15:12] <thomas> wtay: ok, the info clearly shows that pipeline1:channel1 and adderel:sink0 get connected...
[15:12] <thomas> .. what else can I try ? Should I add queues to some of them ?
[15:13] <thomas> To make it run threaded ?
[15:13] <thomas> ... or can I print the scheduled plan somehow ?
[15:13] Nick change: taazzzz -> taaz
[15:18] <wtay> thomas: I tried everything yesterday with gstplay (threads/queues etc)
[15:18] <wtay> thomas: I couldn't get any of them to work with INCSCHED1
[15:19] <thomas> wtay: so it's just that incsched doesn't work like it should then for now ?
[15:19] <thomas> I should hold off working on it then
[15:22] ajmitch (ajmitch at p58-max11.dun.ihug.co.nz) left irc: Ping timeout for ajmitch[p58-max11.dun.ihug.co.nz]
[15:24] <taaz> grr...  debian being goofy.  latest unstable stuff doesnt have a /usr/lib/libdb.so link so libgnome-config detection code is failing at link time...
[15:30] <taaz> but it looks like libdb3-dev is supposed to have it... and a resolved bug report says so too... hmm
[15:31] <wtay> thomas: yeah, Omega is still fixing some things
[15:31] Nick change: wtay -> wtay-away
[15:46] richardb (richard at ixion.tartarus.org) left irc: [BX] Silly wabbit, BitchX is for kids!
[15:47] ajmitch (ajmitch at p35-max5.dun.ihug.co.nz) joined #gstreamer.
[16:48] Ow3n (owen at ti34a80-0313.bb.online.no) left irc: [x]chat
[16:51] Nick change: maYam -> maYam_away
[17:24] dobey (dobey at dreadnought.ximian.com) joined #gstreamer.
[17:37] <taaz> hmm... i suppose i need a colorspace package too since that depends on hermes
[17:39] <thomas> taaz: colorspace is a plugin in the gstreamer hierarchy, no ?
[17:40] <taaz> huh?
[17:40] <taaz> plugins/filters/colorspace
[17:44] <thomas> taaz: so if you have Hermes installed, then colorspace should compile and work
[17:45] <taaz> i'm just talking to myself... i'm speaking of making a seperate debian package
[17:48] <taaz> i've got all the plugins that have external lib dependencies in their own packages. and everything else in one big lump.  it's getting rather messy though...  too many packages
[17:48] <thomas> hmmm, yeah.
[17:48] <thomas> gstreamer is getting clunky
[17:48] <Happyfeet> i had problems with my build till i installed hermes, now it works fine, although the media player is not incredible, its cool
[17:48] <thomas> and people haven't even started writing plugins yet
[17:50] <taaz> well, i've got 23 packages right now.  about to add 2 more.  that is just kinda silly.  but if it isn't split up then people will complain that installing gstreamer pulls in >23 other libs when you install it...
[17:51] <thomas> hmmm... an interface like the apache toolbox would be nice
[17:51] <thomas> ever used that ?
[17:51] <taaz> nope
[17:51] <taaz> what is it?
[17:52] <thomas> well, it's a configure and download script...
[17:52] <thomas> ... you run it, select your options and the packages you want
[17:53] <thomas> and it automatically downloads, configures and compiles all these packages
[17:53] <thomas> so you can ask for apache, compiled with php in, some apache modules, and support for mysql
[17:54] <taaz> what does that have to do with debian packaging?
[17:54] <taaz> i'm going to have packages for everything i can
[17:54] <thomas> it doesn't; you use it when you want to start from source
[17:54] <taaz> and users can install whatever they want
[17:55] <taaz> well, if you're starting from source, just apt-get install libfoo-dev and you get support for foo
[17:56] <taaz> or actually... if you're building the debs from source, the package dependencies will force you to download everything
[17:56] <thomas> yeah, debian is probably better in that respect anyway
[17:57] <taaz> oh well.. i'll ask the debian lists how best to manage this monster
[17:57] Nick change: taaz -> taaz-away
[19:00] steveb (steveb at node1ee31.a2000.nl) joined #gstreamer.
[19:10] thomas (thomas at 212-100-172-175.adsl.easynet.be) left irc: [x]chat
[19:15] Zeenix (programmer at host-93.netzone.net.pk) joined #gstreamer.
[19:19] Zeenix (programmer at host-93.netzone.net.pk) left irc: 
[19:20] Zeenix (programmer at host-93.netzone.net.pk) joined #gstreamer.
[19:20] <Zeenix> hi there
[19:25] <Zeenix> what is the prenounciation of 'gstreamer'
[19:25] <Zeenix> ?
[19:25] <dobey> g-reamer
[19:25] <dobey> err
[19:25] <dobey> streamer
[19:26] <Zeenix> i doubt it
[19:27] <Zeenix> do you know the prenounciation of 'gnome'
[19:27] <dobey> yes, what else do gnomes sound like?
[19:27] <Zeenix> its 'gunome'
[19:27] <Zeenix> like its 'gunu' for gnu
[19:28] <dobey> it's a matter of preference
[19:28] <dobey> i say both
[19:28] <dobey> and miguel doesn't smack me for it, so blah
[19:28] <Zeenix> i dont say you are wrong
[19:28] <Zeenix> any way it doesnt matter
[19:29] <dobey> <- works at ximian if you didn't see that
[19:29] <Zeenix> i need to use gstreamer on my server with no X
[19:29] <Zeenix> can i do that
[19:30] <dobey> i would imagine it possible, as only the gui stuff needs X afaik
[19:33] CHW (paul at temple-baptist.com) joined #gstreamer.
[19:36] steveb (steveb at node1ee31.a2000.nl) got netsplit.
[19:36] Zeenix (programmer at host-93.netzone.net.pk) got netsplit.
[19:36] ChiefHighwater (paul at temple-baptist.com) got netsplit.
[19:36] steveb (steveb at node1ee31.a2000.nl) returned to #gstreamer.
[19:37] Nick change: CHW -> ChiefHighwater
[19:37] Possible future nick collision: ChiefHighwater
[19:40] Zeenix (programmer at host-93.netzone.net.pk) returned to #gstreamer.
[19:44] <Zeenix> no help
[19:47] Zeenix (programmer at host-93.netzone.net.pk) left irc: 
[20:30] omega_ (omega at omegacs.net) joined #gstreamer.
[20:35] <steveb> yo
[20:35] <ChiefHighwater> ello
[20:35] <omega_> yo
[20:39] wtay-away (wim at cable-195-162-214-190.upc.chello.be) got netsplit.
[20:39] dobey (dobey at dreadnought.ximian.com) got netsplit.
[20:39] dobey (dobey at dreadnought.ximian.com) returned to #gstreamer.
[20:50] wtay-away (wim at cable-195-162-214-190.upc.chello.be) got lost in the net-split.
[20:56] hadess (hadess at pc121-gui14.cable.ntl.com) joined #gstreamer.
[20:56] <omega_> yo
[20:57] <dobey> yo
[20:57] <hadess> heya
[21:16] Action: omega_ does a make clean;make
[21:20] richardb (richard at ixion.tartarus.org) joined #gstreamer.
[21:21] Action: richardb is watching the most gorgeous sunset from his study window. :)
[21:24] Action: omega_ hands richardb a webcam
[21:24] <richardb> it's faded now. :(
[21:24] <richardb> But was good last night, so maybe tomorrow night too. ;-)
[21:33] wtay (wim at cable-195-162-214-190.upc.chello.be) joined #gstreamer.
[21:33] <omega_> yo
[21:33] <wtay> yo
[21:34] <wtay> I've been playing with mpeg2enc a bit, 34fps encoding rate... not bad
[21:34] <omega_> neat
[21:36] <Happyfeet> i'd like to encode Taxi Driver in mpeg, would the full movie fill more than a gig ?
[21:36] <dobey> uh, depends
[21:36] <wtay> Happyfeet: depends...
[21:38] <omega_> Wim Taymans of the GStreamer project is also working on a GStreamer
[21:38] <omega_> provider for Bonobo-Media.
[21:38] <omega_> how's that going?
[21:38] <hadess> wtay: good perfs, is it optimised ?
[21:38] <Happyfeet> just curious, like to buy a 75 gig drive and fill it with movies on demand
[21:38] <wtay> hadess: very
[21:38] <wtay> omega_: audio play fine, video doesn't
[21:38] <hadess> wtay: asm, mmx and shit ?
[21:39] <wtay> hadess: yes, lots of it
[21:39] <hadess> gah
[21:39] <wtay> jpeg decoding is actually also done in that 38fps number
[21:39] <omega_> hadess: the mpeg2enc author is very interested in libcodec, which means that hopefully sampeg will be very portable to optimized ppc
[21:40] <wtay> when someone write the assembler that is...
[21:40] <hadess> we just need to find an altivec, g3 asm guru
[21:40] <omega_> yup
[21:41] <hadess> i'm sure ridgerun would be interested as ppc are quite used in embedded devices
[21:41] <hadess> i some ppc optimised mpeg2dec sources here as well
[21:41] <hadess> +have
[21:41] <dobey> ooh
[21:41] <dobey> if xine would have only worked
[21:42] <hadess> dobey: the ones from paulus
[21:42] <dobey> heh
[21:42] <omega_> hadess: can you send them to me?
[21:42] <hadess> omega_: sure
[21:42] <hadess> omega_: dcc ?
[21:42] <omega_> mail
[21:42] Action: omega_ is masq'd
[21:44] Nick change: ajmitch -> aj_uni
[21:44] Action: ChiefHighwater is masq'd 8-]
[21:44] Nick change: maYam_away -> maYam
[21:44] Action: wtay is masq'd
[21:44] <omega_> yo
[21:44] <maYam> hey omega
[21:45] <omega_> ok, this is turning into a survey <g>
[21:45] <dobey> hi
[21:45] <maYam> oooh, my dearest dobey!
[21:45] <wtay> maYam: you're mas'd too
[21:45] Action: ChiefHighwater unmasq'd :-] <gasp>
[21:45] <dobey> hey baby
[21:45] Action: hadess unmasq'd because he has only one computer
[21:45] <maYam> first time i hear about it
[21:45] <hadess> omega_: sent
[21:46] <hadess> hey maYam
[21:46] Action: ChiefHighwater remasq'd 8-] <whew>
[21:46] <maYam> hi hadess
[21:46] Action: ChiefHighwater is having Phantom of the Opera delusions
[21:47] steveb (steveb at node1ee31.a2000.nl) left irc: [x]chat
[21:47] <wtay> pff.. bonobo-media-2.0.tar.gz is invalid
[21:47] <hadess> somebody knows what intl/po2tbl.sed.in is ?
[21:48] <hadess> my configure keeps whining about it not being there
[21:48] <omega_> I would guess it comes from gnome cvs intl/
[21:48] <omega_> or you have to create one
[21:48] <hadess> omega_: does yeah
[21:48] <hadess> the intl/ dir is created automatically when you run autogen
[21:49] <omega_> hmm
[21:53] <hadess> switched to use AM_GNU_GETTEXT instead of AM_GNOME_GETTEXT
[21:53] <omega_> as soon as you figure out how to get gettext to build properly with automake and stuff, can you look into finally adding that to gstreamer?
[21:53] <omega_> or should we ping Kenneth and jump straight to xml-i18n ?
[21:55] <hadess> ping kenneth in any case =)
[21:55] <omega_> true
[21:56] <omega_> hadess: if I can get altivec into libcodec, can I use your imac again to test libmpeg ?
[21:56] <hadess> no altivec here
[21:57] <omega_> oh, right ;-(
[21:57] <omega_> is the idct also altivec?
[21:57] <hadess> i don't know :/
[21:57] Uraeus (cschalle at c224s9h5.upc.chello.no) joined #gstreamer.
[21:57] <omega_> yo
[21:57] <hadess> hey Uraeus
[21:58] <maYam> hi Uraeus
[21:58] <wtay> hi Uraeus
[21:58] <Uraeus> hi happy people not covered in paint like I am :)
[21:58] <maYam> woohoo an Uraeus story!
[21:58] <omega_> horse dumped you into some cans of paint?
[21:59] <Uraeus> :), nothing really exiting just standing under a boat painting all day :)
[21:59] <omega_> doh
[21:59] <Uraeus> but since the paint is bloodred I have gotten that ultracool streetfighter look 
[22:00] <omega_> oooh, take a picture <g>
[22:01] <Uraeus> heh, well I have started to clean it of, but I the only thing I have useable to remove it is aftershave so I smell damned good now :)
[22:01] <omega_> either that or toxic
[22:02] <Uraeus> could be that too, I used a old bottle of Minotaure aftershave so it isn't exactly fresh
[22:02] <dobey> use mineral spirits
[22:02] <Uraeus> mineral spirits? for guys who like to smell like a rock?
[22:03] <hadess> heh
[22:03] <omega_> doh
[22:03] <dobey> no, to get the paint off
[22:03] <omega_> *then* take a bath in the aftershave
[22:03] <dobey> make sure you don't have any open cuts though
[22:03] <dobey> ;-)
[22:03] <omega_> oooh, ouch
[22:05] <Uraeus> well, enough aftershave for today, already smell like I been gay raped by a group of yuppies 
[22:05] <maYam> lol
[22:05] <omega_> eesh
[22:05] Action: omega_ is getting close to finishing a long email on incsched concepts
[22:06] Action: maYam has the solution, do not remove the paint
[22:06] <maYam> it's easy and cheap
[22:06] <omega_> but he then has to let it dry, which takes time
[22:07] <Uraeus> hmm, well I guess I could try to impress the guys at work saying that I dating this excentric body-painting girl
[22:07] <omega_> doh
[22:09] <hadess> Uraeus: muhaha
[22:09] <richardb> Hmm, has anyone taken a look at CSL yet?
[22:09] <omega_> nope
[22:09] <maYam> heheh
[22:10] <Uraeus> richardb: only its creators :)
[22:10] <richardb> Hmm - I've now glanced at one file, too. ;-)
[22:11] <hadess> what is CSL ?
[22:12] <richardb> http://www.arts-project.org/doc/csl-0.1.1.html
[22:12] <Uraeus> Castrated Sound Layer
[22:12] <omega_> hehehe
[22:12] <hadess> heh
[22:12] <richardb> Hey: you should be rude _after_ you've looked at it.
[22:12] <hadess> yeah, ok, didn't remember the name 
[22:12] <richardb> Oh - I mean, _only_ after you've looked at it. ;-)
[22:13] Action: hadess looks at http://www.apple.com/ibook/ some more
[22:13] Action: omega_ hands hadess a napkin to wipe up the drool
[22:13] <hadess> omega_: thanks
[22:14] <omega_> just don't short out your keyboard
[22:14] <Uraeus> richardb: if I where to put that restriction on myself I could never be rude, and my personality would be hampered :)
[22:14] <hadess> i want to replace my imac with that
[22:14] <hadess> or maybe in addition, depends if i get to keep my job :/
[22:15] <Uraeus> hadess: did the Sony Ericsson deal affect you in any way?
[22:15] <hadess> Uraeus: no, the "ericsson cuts staff" has
[22:15] <omega_> did sony buy ericcson or something?
[22:15] <Uraeus> hadess: I know about that, but the Sony deal hasn't changed anything I recon
[22:15] <hadess> Uraeus: nope
[22:15] <Uraeus> omega_: they merged their cell phone divisions or sometihng
[22:16] <omega_> hmm
[22:16] <omega_> under who's name?
[22:16] <hadess> sony didn't make any mobile phones before
[22:16] <hadess> don't know
[22:17] <Uraeus> gstreamer compilation fails with: Making all in mono2stereo
[22:17] <omega_> hadess: do you know who the author of this mpeg2dec ppc stuff is?
[22:17] <Uraeus> '/bin/sh: cd: mono2stereo: No such file or directory'
[22:17] <hadess> omega_: paulus mackerras
[22:17] <omega_> ok
[22:17] <hadess> (sp?)
[22:17] <hadess> you know him i suppose
[22:17] <omega_> nope, but he may be needing a job soon ;-(
[22:17] <omega_> paulus at linuxcare.com
[22:18] <hadess> yeah :/
[22:18] Action: Uraeus points kindly to his compilation problem 
[22:18] Action: omega_ is having trouble figurig out where the altivec intrinsics come from...
[22:19] <omega_> Uraeus: an actual error message helps <g.
[22:19] <hadess> there is no altivec in there afaik
[22:19] <omega_> hadess: oh?
[22:19] <Uraeus> '/bin/sh: cd: mono2stereo: No such file or directory'
[22:19] <omega_> motion_comp_altivec.c
[22:19] <omega_> Uraeus: did you update with -d ?
[22:19] <Uraeus> uh, no
[22:19] <hadess> omega_: that's optimised for the decoder of the TiVo
[22:19] <omega_> cvs update -dP
[22:19] <Uraeus> ok
[22:19] <hadess> omega_: ok, there's altivec :P
[22:20] <omega_> hadess: TiVo has hardware to decode
[22:20] <richardb> Uraeus: you'll then need to re-run autogen too, probably...
[22:21] <Uraeus> ok
[22:21] <omega_> after you apply richardb's patch to /usr/bin/automake <g>
[22:22] <hadess> omega_: yeah, the software to drive it has been removed from these sources
[22:22] <omega_> oh?
[22:22] <omega_> but it does exist? is it available?
[22:23] <hadess> under NDA as i toldyou
[22:23] <omega_> ah
[22:23] <hadess> +\ 
[22:23] <hadess> linuxcare has done work on the TiVo
[22:24] <omega_> ah
[22:24] <hadess> he usually hangs out on #mklinux so you can get to talk to him directly if you want
[22:25] <omega_> ok
[22:29] <Uraeus> ok, gstreamer still fails, this time in mono2stereo where is says 'no rule to make all'
[22:30] <omega_> hmm, someone didn't add everything to cvs
[22:30] <Uraeus> not me
[22:31] <omega_> hmm
[22:31] <omega_> is there a mono2stereo/Makefile ?
[22:32] <Uraeus> just .am and .in
[22:32] <omega_> that's a problem
[22:32] <omega_> you ran autogen.sh ?
[22:32] <Uraeus> yes
[22:32] <Uraeus> after I updated with yout cvs command I reran autigen.sh
[22:32] <omega_> you updated configure.in to latest HEAD ?
[22:32] <omega_> hmmmm
[22:32] <richardb> Hmm; its in configure.in
[22:33] <omega_> right, that's why /me is confused
[22:33] <richardb> I think it worked for me...
[22:33] Action: hadess is away: f00d
[22:33] <richardb> yes.
[22:33] <Uraeus> omega_: uh? don't understand do I need to do something special to get latest configure.in?
[22:33] <richardb> just cvs update in home dir.
[22:33] <omega_> not if you did a cvs update -dP
[22:33] <Uraeus> I did
[22:33] <omega_> root of working copy, that is
[22:33] <richardb> s/home dir/top dir/
[22:34] <omega_> try update -dPA
[22:34] <Uraeus> ok
[22:35] Nick change: taaz-away -> taaz
[22:36] <taaz> hmm?  hadess?  did you say you had ppc mpeg2dec code?
[22:36] <richardb> csl summary from a quick check: it's ick, but better than artsd...
[22:36] <richardb> the interface also looks like it could be implemented in a threadsafe manner...
[22:36] <richardb> though I understand that the arts implementation wouldn't be.
[22:36] <omega_> 'could be' ?
[22:36] <taaz> why not just use alsa as your 'common' layer?
[22:37] <omega_> because it's got its own problems
[22:39] <richardb> A virtualised device does make sense in some situations...
[22:39] <richardb> Got to go.
[22:39] richardb (richard at ixion.tartarus.org) left irc: [BX] Reserve your copy of BitchX-1.0c16 for the Commodore 64 today!
[22:49] Action: omega_ just ordered a DSC-C300 usb webcam and an 8-port 10/100 switch to replace his completely fried 5-port switch
[22:50] <taaz> DSC-C300?  not more ibots? ;)
[22:50] <omega_> no, I have one of those, I want a usb cam too
[22:51] Action: omega_ wants to eventually hook up a usb cam, a 1394 cam, and two DV camcorders to one machine and show them all simultaneously with gstreamer <g>
[22:51] <taaz> i gotta try mine with 2.4.4... see if it crashes the kernel less
[22:51] <omega_> or even mix them
[22:51] <omega_> you have a C300 ?
[22:51] <omega_> dichro said that the latest driver (not in the kernel) works better
[22:52] <taaz> my webcam cost $25 or something after rebate... cheap stuff.  quality sucks.  doesn't work worth a darn in low light...
[22:52] <omega_> ov511 ?
[22:52] <taaz> i forget what it is...
[22:52] <taaz> i haven't used it much because of that kernel crashing nonsense
[22:53] <omega_> gstreamer's v4lsrc is much harder on the ov511 driver than anything else, it triggers oopses a lot
[22:55] <hadess> omega_: that's good you have one, you know how it feels for me then :/
[22:55] <omega_> heh
[22:55] <omega_> I'll have it by friday, at least (I hope)
[22:56] <hadess> taaz: i have some code from paulus
[22:56] <omega_> hrm, I had an idea
[22:57] <omega_> tomshardware claims that flask has had problems with audio level being too low
[22:57] <omega_> afaict, ac3dec is the culprit
[22:58] <hadess> flask ?
[22:58] <omega_> FlasKMPEG, the DVD-ripper's tool on win32
[22:59] <hadess> ok
[23:00] <taaz> hmm not the livid ac3dec?
[23:00] <omega_> wtay: the autoplugger is still not working right in incsched
[23:00] <omega_> the ac3dec we have in gstreamer sounds very quiet
[23:00] <taaz> there is another windoze project and domain name(s) of the same name
[23:00] <omega_> and flask uses a derivitive of ac3dec as well
[23:01] <omega_> nope, it's the same one
[23:01] <omega_> flask uses livid's ac3dec
[23:01] <omega_> unoptimized
[23:02] <taaz> i see... and do you think they have made any modifications?
[23:02] <omega_> probably
[23:02] <taaz> cause they sure haven't bothered to post patches to the livid lists
[23:03] <omega_> nope
[23:04] <omega_> the changes are mostly indentation style
[23:04] <omega_> but there are some code changes, nothing major
[23:04] <omega_> but this is just looking at one file
[23:04] <omega_> go.to/flask/ iirc
[23:06] <taaz> i've got other things to work on ;)
[23:07] <taaz> i think i'll be up to 25 (!) debian gstreamer packages
[23:07] <omega_> eek
[23:07] <taaz> which is borderline unreasonable
[23:08] <taaz> i'm not sure which side of that border ;)
[23:08] <omega_> quite
[23:08] <omega_> but then there are so many plugins with external dependencies, it's unavoidable
[23:08] <taaz> ?  it's completely avoidable.  just lump it all together and say "deal with it"
[23:09] <taaz> the libs themselves are pretty small...
[23:09] <taaz> and in practice people will probably install everything anyway
[23:09] <taaz> but, uh, i'm targeting embedded debian distros that only want to install a few pcakges... yeah...
[23:10] <omega_> heh
[23:11] <taaz> i need to make artsd and colorspace packges... i think that will bring me up-to-date
[23:12] <omega_> and dvdec and dv1394src ?
[23:12] <taaz> it seems questionable for things like colorspace... seems just about any video app will use it.
[23:12] <omega_> well, the problem is that it depends on Hermes, which I don't like much
[23:12] <omega_> but may be unavoidable
[23:12] <taaz> hmm...  i haven't done -all- the plugins yet.  some depend on libs not in debian. 
[23:12] <omega_> oops
[23:13] Uraeus (cschalle at c224s9h5.upc.chello.no) left irc: syntax error - user imploded
[23:14] <taaz> along those lines, lame isn't in debian so that plugin will be unofficial.  mpeg2dec not packaged so that's more work for me... since i really want that one in there
[23:17] Nick change: omega_ -> omega_lunch
[23:17] <hadess> taaz: thinking of packaging lame as well ?
[23:18] <hadess> taaz: given that there's a debian/ dir in the CVS, that shouldn't be too hard...
[23:20] <omega_lunch> if it's anything like the spec file in cvs, it's an utter mess
[23:21] Nick change: omega_lunch -> omega_
[23:25] _gst_newt_ joined #gstreamer.
[23:25] omega_ (omega at omegacs.net) joined #gstreamer.
[23:26] maYam (mayam at cable-195-162-214-190.upc.chello.be) joined #gstreamer.
[23:26] ChiefHighwater (paul at temple-baptist.com) joined #gstreamer.
[23:27] taaz (dlehn at joined #gstreamer.
[23:27] matth_ (matth at qwest.dsplinux.net) joined #gstreamer.
[23:27] aj_uni (ajmitch at p35-max5.dun.ihug.co.nz) joined #gstreamer.
[23:28] Nick change: aj_uni -> ajmitch
[23:30] wtay (wim at cable-195-162-214-190.upc.chello.be) joined #gstreamer.
[23:35] Nick change: ajmitch -> aj_uni
[23:37] <taaz> hadess: no, it's a free software/patent issue thing.  i think they have chosen to not have lame in debian at all
[23:37] <omega_> I thought lame was gpl'd ?
[23:37] <hadess> patent
[23:38] <hadess> not copyright
[23:38] <omega_> oh, right
[23:38] <hadess> taaz: even an unofficial one ?
[23:38] <taaz> uh.. anyone is welcome to build the package...
[23:39] <taaz> it's irrelevent anyway... eveyone should use vorbis ;)
[23:40] Action: wtay is building INCSCHED again
[23:40] Action: taaz waits for the "but my hardware mp3 player doesn't do that" whine ;)
[23:40] <hadess> except that there's no vorbis on my ...
[23:40] <hadess> ok
[23:40] <hadess> heh :P
[23:40] <omega_> grrrr, I need to rotate a .ps before i convert it to a .pdf
[23:40] ChiefHighwater (paul at temple-baptist.com) left irc: Read error to ChiefHighwater[temple-baptist.com]: Connection reset by peer
[23:41] Action: hadess needs to install mpeg2dec
[23:41] <hadess> which version is the recommended one ?
[23:42] <hadess> -m+d
[23:43] <wtay> not much choice: 0.2.0
[23:44] <hadess> ok, so it will compile with that you think ?
[23:45] <wtay> yes, it should (it does here)
[23:45] <omega_> works for me
[23:45] <hadess> cool
[23:45] <hadess> will try to play some vobs with that 
[23:46] ChiefHighwater (paul at temple-baptist.com) joined #gstreamer.
[23:47] <hadess> whaoo, compiles in under a minute
[23:51] <taaz> under a minute?  let me guess... you're not talking about gstreamer ;)
[23:52] <hadess> mpeg2dec :)
[00:00] --- Wed May  2 2001
[00:12] <matth_> omega_: what's up?
[00:12] <omega_> getting ready to write the last section (iteration) of the big explanation of incsched
[00:12] <omega_> then do another HEAD->INCSCHED merge
[00:12] <matth_> what's changed in HEAD?
[00:12] <omega_> then maybe go ahead with the INCSCHED->HEAD merge
[00:12] <wtay> omega_: I see strange things...
[00:12] <omega_> wtay: I see dead code...
[00:13] <omega_> (to 'quote' a recent movie)
[00:13] <wtay> #0  0x8108bfd in ?? ()
[00:13] <wtay> #1  0x4002718b in gst_element_set_state (element=0x80f0cb0, state=GST_STATE_READY) at gstelement.c:762
[00:13] <wtay> #2  0x40020af7 in gst_bin_change_state (element=0x80f0900) at gstbin.c:396
[00:13] <wtay> #3  0x4003db22 in gst_thread_main_loop (arg=0x80f0900) at gstthread.c:444
[00:13] <omega_> interesting
[00:13] <wtay> segfaults all over the place in stupid places (gst_buffer_destroy) etc..
[00:13] <omega_> did you do a complete make clean;make ?
[00:13] <omega_> I had to do that for some reason
[00:14] <wtay> ok, doinf that now, mpeg2parse4 played perfect though
[00:14] <omega_> something changed that caused the linker and compiler to have a cow
[00:14] <wtay> strange
[00:15] <hadess> ** ERROR **: file gstprops.c: line 216 (gst_props_newv): should not be reached <- again
[00:15] <omega_> that'd be wim's code
[00:15] <hadess> every time...
[00:15] <hadess> that's about 2 weeks that i can't even run the code :/
[00:15] <wtay> when?
[00:15] <omega_> uh, that's some strange code there...
[00:16] <hadess> gst-reg
[00:16] <omega_> default: lock, chunk_free, unlock, g_assert_not_reached
[00:16] <omega_> hadess: HEAD or INCSCHED ?
[00:17] <hadess> HEAD is default, righ ?
[00:17] <omega_> yes
[00:17] <wtay> strange but perfectly legal code..
[00:17] <hadess> HEAD then
[00:17] <omega_> wtay: but why are you doing stuff, then asserting that the next line is never reached?
[00:17] <wtay> hadess: loading what plugin?
[00:18] <hadess> INFO(16035:-1):gst_plugin_load_absolute:368: loading plugin "/home/hadess/src/gstreamer/./gst/elements/.libs/libgstelements.so"...
[00:18] <wtay> omega_: hmm yeah...
[00:18] <wtay> omega_: we don't cause a leak before we exit the app right :-)
[00:18] <omega_> um, right <g>
[00:19] Action: omega_ thwacks wtay with a large trout
[00:19] <hadess> heh
[00:19] <wtay> I think I wanted to fail gracefully but decided to assert in the end...
[00:19] <omega_> doh
[00:21] <wtay> hadess: since when do you have that problem?
[00:21] <hadess> after guadec... when did i send you mp3mad ?
[00:21] <hadess> it was already not working at that time
[00:26] <wtay> must be the props API changes using va_args
[00:26] Nick change: matth_ -> matth-vacation
[00:26] <matth-vacation> :-)
[00:26] <wtay> grr
[00:26] Action: matth-vacation goes mountain biking to moab, utah
[00:29] <wtay> hadess: I'll try to fail gracefully and g_print some stuff..
[00:29] <hadess> ok
[00:30] <omega_> incsched info sent
[00:31] <wtay> omega_: still the same crashes after make clean...
[00:31] <omega_> um
[00:31] <omega_> running what?
[00:31] <wtay> gstplay
[00:31] <omega_> on a..
[00:31] <wtay> in typefind
[00:31] <omega_> what kind of file?
[00:32] <wtay> mpeg1
[00:32] <omega_> system?
[00:32] <wtay> pipeline (disksrc -> typefind)
[00:32] <wtay> yup
[00:32] <wtay> faults in gst_buffer_unref
[00:32] <wtay> well _destroy acually
[00:32] <omega_> I get something entirely different
[00:32] <wtay> could be I have some pending changes
[00:32] <omega_> ** CRITICAL **: file gstplay.c: line 131 (gst_play_init): assertion `priv->audio_play != NULL' failed.
[00:33] <wtay> osssink or esdsink?
[00:33] <omega_> dunno, but no difference whether esd is running or not
[00:33] <omega_> ossink
[00:33] <omega_> changing to esdsink
[00:34] <omega_> with the knowledge that the audio will suck
[00:34] <wtay> haddes can you test the CVS gstprops.c
[00:35] <wtay> omega_: care if I checkin the mods I made?
[00:35] <omega_> what do they change?
[00:35] <wtay> pipeline setup
[00:35] <wtay> that was screwed
[00:35] <omega_> in autoplug?
[00:35] <wtay> yup
[00:35] <wtay> in gstplay
[00:35] <omega_> ok
[00:36] <taaz> what is gstossaudio? (too lazy to look at it)
[00:36] <omega_> the osssink and osssrc plugin
[00:36] <taaz> ok
[00:37] <wtay> omega_: in CVS
[00:37] <omega_> only in gstplay?
[00:37] <wtay> omega_: yes, for now
[00:37] <hadess> omega_: can you send an email saying what it changes for application developers ?
[00:38] <omega_> yeah, that's next, as soon as we all figure it out
[00:38] <hadess> ok, cool
[00:38] <omega_> Gtk-WARNING **: gtk_signal_connect(): could not find signal "handoff" in the `GstEsdsink' class ancestry
[00:39] <wtay> ok, ignore that one...
[00:39] thomas (thomas at adsl-64926.turboline.skynet.be) joined #gstreamer.
[00:39] <wtay> hi
[00:39] <thomas> hi
[00:39] <omega_> I get a fault at gstscheduler.c:922
[00:39] <wtay> hmm
[00:39] <thomas> omega_: I'm readin your incsched explanation now, but I'll have to continue tomorrow...
[00:39] <omega_> ok
[00:40] <thomas> ... you think I will be ok if I implement the mixer tomorrow the way you describe ?
[00:40] <omega_> which way?
[00:40] <thomas> I mean, will it solve the problem I'm having today with the cothread switching ?
[00:40] Action: wtay is going to post the autoplug2 explanation to -devel
[00:40] <omega_> thomas: it should, I think
[00:40] <thomas> Right now I just took the code as it worked in head and tried to delay requesting a new pad...
[00:40] <thomas> ... but that didn't work in incsched, so...
[00:40] <wtay> thomas: it might work now...
[00:41] <thomas> wtay: as in right this instant ?
[00:41] <thomas> ok, then I'll check quickly before going to bed
[00:41] <omega_> basically, the goal by the end of the next weekend is for everyone's test programs to be updated to incsched1, and a good doc describing what you can and can't do with incsched
[00:42] <thomas> omega_: sounds good.  you move fast.
[00:42] <omega_> we're starting to get schedules at rr, so... <g>
[00:43] <thomas> schedules always help in the transition from imagination to reality ;)
[00:43] <thomas> one more thing : can I add the level plugin to cvs or should I wait for that ?
[00:44] <omega_> um, I suppose you can, yeah, since we'll need to finally do our complete cvs reorg soon anyway
[00:44] <thomas> well, it's your call.  I mean, it's really basic anyway.
[00:44] <thomas> If you'd rather hold off until incsched is merged then that's fine.
[00:44] <omega_> yeah, go ahead
[00:45] <thomas> ok... how do I create a dir ? with cvs import ?
[00:45] <omega_> there will be a freeze during the double merge of incsched, but that shouldn't last more than an hour
[00:45] <omega_> mkdir abc;cvs add abc
[00:45] <omega_> the populate it, and cvs add the files
[00:45] <omega_> don't try to cvs add a populated dir, it doesn't work very well
[00:46] <thomas> hmmm... mixer gave a segfault in the typefinder...
[00:47] <omega_> bleagh
[00:47] <thomas> oh well... I'll check it tomorrow
[00:47] <wtay> weird stuff..
[00:48] Action: omega_ is an idiot
[00:49] <wtay> hmm?
[00:49] <omega_> I broke the code in the scheduler
[00:49] <omega_> fixing
[00:49] <ChiefHighwater> details, details
[00:49] <thomas> ok, night all
[00:49] <omega_> g'night
[00:49] <ChiefHighwater> gnite 8-]
[00:49] <wtay> thomas: cya
[00:49] thomas (thomas at adsl-64926.turboline.skynet.be) left irc: Client Exiting
[00:49] <wtay> omega_: related to my misterious crashes?
[00:50] <omega_> doubtful
[00:50] <omega_> but possible
[00:50] <wtay> mem corruption?
[00:50] <omega_> maybe
[00:50] Action: omega_ attempts to fix his fix of a fix
[00:51] <omega_> ok, that's fixed, committing
[00:51] <omega_> but it's still hardlocked somewhere, bt'ing
[00:52] Action: wtay runs apt-get update for the 5th time today to kill time...
[00:52] <omega_> interlock code ;-(
[00:52] Action: omega_ runs red-carpet for the first time in a week to kill time...
[00:53] <omega_> maybe I can finally upgrade to Ximian GNOME 1.4
[00:53] <wtay> heh, I got Xfree 4.0.3 yesterday...
[00:54] <omega_> this is a little faster than what I was getting a week ago
[00:54] <omega_> 51KB/sec vs. 2KB/sec
[00:54] <wtay> slightly :)
[00:55] <ChiefHighwater> ouch
[00:56] <omega_> I'm gonna go fix up the DEBUG system to look like the INFO system really quick
[00:56] <omega_> else I won't be able to follow the output from gstmediaplay <g>
[00:57] <wtay> ok, no crash now, just a lockup
[00:58] <omega_> yeah, in gst_thread_main_loop() ?
[00:58] <wtay> no
[00:59] <omega_> in gst_thread_change_state then?
[00:59] <wtay> looks like a queue is never emptied
[00:59] <omega_> hmm
[00:59] <omega_> matth changed that stuff recently, it would be worth figuring out what he did
[01:00] <omega_> does autoplug do the same thing?
[01:00] <wtay> cothread: can't switch to NULL cothread!
[01:00] Last message repeated 1 time(s).
[01:00] <omega_> great
[01:01] <wtay> DEBUG(24568: 0)gst_schedule_iterate:1305: there are 1 elements in this chain
[01:01] <wtay> DEBUG(24568: 0)gst_schedule_iterate:1311: entry "queue_mad" is DECOUPLED, skipping
[01:01] <wtay> DEBUG(24568: 0)gst_schedule_iterate:1318: set COTHREAD_STOPPING flag on "queue_mad"(@0x80e4d58)
[01:01] <wtay> cothread: can't switch to NULL cothread!
[01:01] <wtay> DEBUG(24568: 0)gst_schedule_iterate:1326: cothread switch ended or interrupted
[01:01] <wtay> DEBUG(24568: 0)gst_schedule_iterate:1301: starting iteration via cothreads
[01:01] <wtay> not sure it that helps..
[01:01] <omega_> not right now, lemme finish the debug stuff first
[01:02] Nick change: maYam -> maYam_sleep
[01:02] <maYam_sleep> wtay.. uh..  don't want to spoil it, but you have to go to work tomorrow..
[01:03] <wtay> maYam_sleep: :(
[01:03] Action: maYam_sleep 's efforts never help
[01:04] <wtay> omega_: gstbuffer.c line 451, newbuf->data = (gchar *)g_malloc (buffer->4data); fixing to size
[01:04] <omega_> did you add the red?
[01:04] <wtay> yes
[01:05] <omega_> oops
[01:05] <omega_> I thought I fixed that
[01:05] <omega_> check HEAD?
[01:05] <omega_> oops
[01:06] Action: wtay found the bug by looking at the compiler warnings
[01:06] <omega_> ah
[01:06] <omega_> yeah, lclint has a whole herd of cows when I run it on gst/*.[ch]
[01:07] <wtay> interesting ones?
[01:07] <omega_> I didn't look, it was too much
[01:08] <omega_> mostly nit-picking, hence the name lint...
[01:08] <wtay> heh, well: gst/gstinfo.h: undocumented I/O error
[01:09] <omega_> *undocumented* I/O error???
[01:09] <wtay> as if there are docs...
[01:09] <omega_> hah
[01:13] <wtay> great, the new vorbis doesn't like to play my old vorbis file...
[01:13] <omega_> neat
[01:13] <hadess> huh, it's supposed to...
[01:13] <wtay> I even tried ogg123 on it..
[01:14] <hadess> bad bad
[01:15] <wtay> looks like queues cause mayor headaches for the scheduler...
[01:15] <wtay> in some way..
[01:15] <wtay> I'm off to bed now, cya all..
[01:15] Nick change: wtay -> wtay-zZz
[01:38] Nick change: hadess -> largo_
[01:38] Nick change: largo_ -> hadess
[01:43] <ChiefHighwater> l8rz y'all   8-]
[01:43] ChiefHighwater (paul at temple-baptist.com) left irc: 
[01:50] hadess (hadess at pc121-gui14.cable.ntl.com) left irc: mooooh!
[05:19] Nick change: taaz -> taazzzz

More information about the gstreamer-devel mailing list