[Ffmpeg-devel-irc] ffmpeg-devel.log.20130222

[burek]

[00:08] <cone-48> ffmpeg.git 03Michael Niedermayer 07master:29436c1ab2c5: libavcodec/mips/aacsbr_mips: fix ;;
[03:01] <cone-48> ffmpeg.git 03Michael Niedermayer 07master:b16bf913d8ed: dxva2: move _WIN32_WINNT handling to public header
[04:43] <tonsofpcs> just ran ffprobe -show_packets on a .ts and it is showing a dts = pts on all packets that don't have a dts.  Why is this?  (it's an ATSC TS if that matters)
[05:42] <llogan> i'm wondering if "qualification tasks" are an impediment to getting shit done for gsoc.
[06:59] <av500> by experience, a lot of "shit" has been done
[08:33] <highgod> Hi,Stefano Sabatini,I have re-submit the opencl patch
[11:01] <cone-583> ffmpeg.git 03Matt Wolenetz 07master:311443f6c7eb: x86: h264: Don't use redzone in AVX h264_deblock on Win64
[11:01] <cone-583> ffmpeg.git 03Michael Niedermayer 07master:db05f7ab1626: Merge remote-tracking branch 'qatar/master'
[13:18] Action: durandal_1707 looking at drawutils, and what to change for >8 bit support ; result = headache
[13:40] <Compn> (Remove the sws project, its mentor thinks it is pointless and a complete rewrite is needed instead.)
[13:40] <Compn> snort, rewrite of swscale :D
[13:44] <durandal_1707> hmm how to know what number of threads are appear when calling execute?
[13:48] <durandal_1707> i could allocate 16 * height bytes in worst case
[13:49] <durandal_1707> because each image scan line could be decoded in separate thread
[13:50] <durandal_1707> but having 8000 thread jobs may be too much
[13:57] <durandal_1707> ught there is memory leak in lavf/mxg
[14:42] <cone-449> ffmpeg.git 03Paul B Mahol 07master:a72b41035cc4: mxg: do not leak mxg->buffer
[14:52] <cone-449> ffmpeg.git 03Michael Niedermayer 07master:21f68528e39d: aacdec: dont pop random configurations
[14:55] <cone-449> ffmpeg.git 03Paul B Mahol 07master:c46400ddecab: cafenc: do not leak caf->pkt_sizes
[15:38] <tclarke> hi, I'm trying to write a custom muxer which write raw frames (using the rawvideo codec) to a circular framebuffer in a memmapped file...the problem I have is that ffmpeg is opening the file I specify and performing flush's, etc. so the second time my write_packet function is called the mmap section seems to be overritten with 0's
[15:39] <tclarke> how can I tell ffmpeg that my AVOutputFormat will handle it's own I/O and an avio context should not be opened?
[15:41] <Compn> sounds complicated
[15:43] <tclarke> you have a better solution?
[15:43] <ubitux> tclarke: maybe look how it's done in the segment muxer
[15:44] <ubitux> or lavf/hlsenc
[15:44] <tclarke> ok, I'll look there, thanks for the pointers
[16:31] <Compn> j-b : could just do what winamp does for interfaces , put it in html , make user use his web browser to see bittorrent status :)
[16:31] <Compn> for bittorrent playback thing
[17:27] <creep> hi
[17:27] <creep> what kind of music do you consume ? 
[17:27] <creep> Flyleaf - i'm so sick, Fully Alive http://www.youtube.com/watch?v=iWIADZKU9dw  hackers used to listen stuff like this on films
[17:32] <gnafu> Cool story, bro.
[17:44] <Compn> hey hes back
[17:45] <mystica555> [libvorbis @ 0x119c8e0] Que input is backward in time
[17:45] <mystica555> ...que hora es? >.>
[17:46] <Compn> is that a libvorbis error or our error ?
[17:47] <mystica555> im not entirely sure
[17:48] <mystica555> it visibly appears similar (with a teal background) to 
[17:48] <mystica555> [libx264 @ 0x11b0600] frame I:46    Avg QP:25.17  size: 26041
[17:48] <mystica555> que... was brown
[17:48] <mystica555> libvorbis teal
[17:50] <Compn> i dont see that error message in the code.
[17:50] <mystica555> then its probably in libvorbis.
[17:50] <Compn> no
[17:50] <Compn> its in libfaac to
[17:50] <Compn> somehow
[17:50] <Compn> lets see where it comes from...
[17:51] <mystica555> hmm
[17:51] <creep> nah
[17:52] <creep> spam? :P
[17:52] <mystica555> /usr/lib64/libavcodec.so.54.59.100: Que input is backward in time
[17:52] <mystica555> hm
[17:52] <mystica555> says strings
[17:52] <mystica555> so definitely there, somewhere.
[17:53] <Compn> audio_frame_queue.c:            av_log(afq->avctx, AV_LOG_WARNING, "Queue input is backward in time
[17:53] <Compn> is it somehow eating "ue input" ?
[17:53] <mystica555> likely a newer version patched it
[17:53] <mystica555> ive maybe not updated in a few months
[17:53] <Compn> maybe its just a typo
[17:53] <Compn> que > queue
[17:54] <Compn> english isnt the first language of many devs in the project 
[17:54] <Compn> and there are multiple ways to spell the same english words :\
[17:54] <mystica555> well yes im sure it is a typo
[17:54] <mystica555> que just isnt one of the right ways
[17:54] <mystica555> either you "cue" something - to advance it to a certain poinbt in time in preperation for playback for example;  or you put something into a queue
[17:55] <mystica555> que is spanish however
[17:55] <mystica555> ;)
[17:56] <Compn> que is slang for queue then :P
[17:56] <Compn> in this case
[17:56] <mystica555> que paso
[17:56] <mystica555> ;)
[17:56] <Compn> vato loco!
[17:58] <Compn> but yeah it appears its been fixed in the meantime
[18:20] <cone-345> ffmpeg.git 03Michael Niedermayer 07master:6f77122bf571: aac: reconfigure output on pop
[18:28] <saste> ubitux: I'm asking because an av_free() would have fail.
[18:28] <saste> what do you mean?
[18:40] <BBB-work> "Description: Not many VP7 samples are in the wild, but no open-source decoder exists although Google has released a specification. Write a decoder that reuses as much as possible of existing FFmpeg code, it is likely that functions of the existing decoders for On2-based formats will be useful." ?
[18:40] <BBB-work> I didn't know google released a vp7 spec?
[18:43] <nevcairiel> all i know of is this: http://multimedia.cx/mirror/VP7_Data_Format_and_Decoder_Overview.pdf
[18:44] <nevcairiel> it says to be used in conjunction with the reference decoder
[18:45] <BBB-work> that is not released by google, that is dropped off a truck
[18:45] <BBB-work> that's not the same thing :)
[18:45] <nevcairiel> since when do we care about that specifically? :d
[18:46] <BBB-work> you're writing a wiki page to give it to a company named google for inclusion in gsoc, right?
[18:47] <BBB-work> that very same company is claimed to have done something it didn't do on that very wiki page
[18:51] <nevcairiel> for all i know, they have released it, whoever added that shall be questioned
[19:21] <ubitux> saste: mmh sorry, it would have fail in some configuration only
[19:21] <saste> ubitux, i still can't understand
[19:22] <saste> how can av_free fail?
[19:22] <ubitux> in case of memalign hack for instance
[19:22] <ubitux> if the allocated buffer is not from av_malloc or similar
[19:22] <ubitux> it would fail
[19:23] <saste> but we are using av_malloc
[19:23] <saste> strdup -> av_strdup should be obviously fixed
[19:23] <ubitux> yes that's the point
[19:23] <ubitux> on a memalign config, the buffer allocated with strdup() would have fail if the pointer was freed with a av_free()
[19:24] <ubitux> memalign-hack* configu
[19:24] <saste> so if i fix strdup -> av_strdup() it will be safe, right?
[19:25] <ubitux> sure
[19:25] <saste> good i'll fix and push then
[19:28] <ubitux> btw, about the [HH:] thing the guy actually sent an updated patch
[19:29] <saste> ubitux, for some reason I got my mails stucked on the server
[19:29] <saste> so they are released just after many hours
[19:30] <saste> the more time i don't fetch mails, the more time they tend to stay stucked
[19:30] <saste> so in my mailbox i don't always have all the mails
[19:31] <saste> it sucks big time, but nothing i can do much about without spending a few hours debugging or nailing tech support
[19:31] <ubitux> :(
[19:34] <ubitux> ffcat
[19:34] <ubitux> fast forward cat
[19:42] <tonsofpcs> I ran ffprobe -show_packets on a .ts and it shows dts = pts on all packets (video and audio) that don't have a dts.  Why is this?  (it's an ATSC TS if that matters)
[19:46] <saste> ffcat -> fast frenzy kittie
[19:55] <creep> http://images.puszta.com/hazi_kifli/Hazi%20kifli%201000.jpg !
[19:55] <Tjoppen> oh mama
[20:01] <cone-345> ffmpeg.git 03Stefano Sabatini 07master:f30979e0954e: lavfi/boxblur: add support to named options
[20:01] <cone-345> ffmpeg.git 03Stefano Sabatini 07master:8902f5b89c16: doc/filters: apply minor spacing fixes to boxblur docs
[20:18] <durandal_1707> michaelni: have anything to add to exr patches? (i have improved version with slice threading)
[20:21] <cone-345> ffmpeg.git 03Paul B Mahol 07master:345689129102: exr: check version too
[20:29] <michaelni> durandal_1707, the patches should be fine
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:88d7259fd246: exr: merge multiple if () blocks
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:1178868683d2: exr: fix overflow check in lineoffset processing
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:cc0bd7841389: exr: support more line orders
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:9c75c24de40d: exr: check if line offset table is complete only once per frame
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:fefd4b0924cc: exr: cosmetics: reindent
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:7a930f310ad2: exr: remove superfluous check
[20:35] <cone-345> ffmpeg.git 03Paul B Mahol 07master:89c77b6b155c: exr: shorter channel buffer names
[20:58] <cone-345> ffmpeg.git 03Michael Niedermayer 07master:2c21d34ea44d: avutil/mem: do a small set of checks for memalign hack before freeing.
[21:30] <code-guru> hi all
[21:31] <code-guru> I can't set mpegts metadata when use hls muxer.
[21:31] <code-guru> for example when I add "-metadata service_provider="Some provider" it doesn't change service_metadata when I use hls m3u8 muxer
[21:31] <code-guru> can anybody help me with that?
[21:33] <llogan> code-guru: sorry. wrong channel. #ffmpeg is for user help.
[21:34] <code-guru> llogan, it seems more of a bug to me ? isn't it ?
[21:53] <ubitux> btw, about GSoC
[21:53] <ubitux> i can suggest some subtitles stuff
[21:54] <ubitux> and i'd be very happy to see improvements in the gif muxer/encoder
[21:54] <ubitux> like non-intra frames, and better colors
[21:56] <llogan> would you be interested in mentoring?
[21:56] <ubitux> subtitles yes, gif maybe
[21:58] <llogan> can you add this to the ideas page?
[21:58] <ubitux> maybe in an hour or two
[21:58] <llogan> or email me the info and i can do it if that will be easier for you.
[21:59] <ubitux> i think i need an account too :p
[22:00] <llogan> but i can't do it until later either...more video editing to do
[22:09] <cone-345> ffmpeg.git 03Malcolm Bechard 07master:239b88c284af: Improve QTRLE encoding performance, no change to output file size/content.
[22:50] <code-guru> https://ffmpeg.org/trac/ffmpeg/ticket/2296
[22:53] <cone-345> ffmpeg.git 03Martin Storsjö 07release/1.1:65bf4c9c4578: mpegvideo: Do REBASE_PICTURE with byte pointers
[22:53] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:0b2b8ab97962: qdm2: check array index before use, fix out of array accesses
[22:53] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:52b18c1fde65: roqvideodec: check dimensions validity
[22:53] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:ab434bf0d051: vqavideo: check chunk sizes before reading chunks
[22:53] <cone-345> ffmpeg.git 03Vicente Jimenez Aguilar 07release/1.1:6626a7df5345: doc: Fix some obsolete references to av* tools as ff* tools
[22:53] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:80ddf7889eec: Merge remote-tracking branch 'qatar/release/9' into release/1.1
[23:11] <ubitux> fun fact: there are around 12k additionnal commits in ffmpeg against l. (without counting merges)
[23:12] <nevcairiel> now do that excluding commits that they ported over
[23:12] <nevcairiel> :D
[23:12] <ubitux> 12k-3 ?
[23:15] <ubitux> ok so anyway
[23:15] <ubitux> i need to mail mike for a multimedia.cx account, right?
[23:15] <wm4> simply subtracting the number of commits shows that ffmpeg has slightly more commits, I think?
[23:15] <wm4> though you'd have to subtract merge commits too, hm
[23:15] <ubitux> i didn't count them
[23:16] <ubitux> otherwise it's 14k or something iirc
[23:16] <ubitux> but TBH i should have count them somehow :)
[23:16] <ubitux> llogan: ping?
[23:21] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:a642be972d48: h264: ensure that get_format() is called when changing format but not otherwise.
[23:21] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:469cb6119386: avcodec_decode_audio4: check got_frame_ptr before handling initial skip
[23:21] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:02d1efdd5b61: h264: check that luma and chroma depth match
[23:21] <cone-345> ffmpeg.git 03Andrea3000 07release/1.1:ef688e742557: matroska: fix missing ,
[23:21] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:705e89d75fde: update for 1.1.3
[23:41] <michaelni> BBB-work, how should the VP7 text on the SOC page be worded ? (i dont know at all where which spec came from)
[23:42] Action: michaelni just found his un/pw for wiki-multimedia again ;)
[23:42] <ubitux> can you find mine if it exists?
[23:44] <michaelni> ubitux, if you had an account try to login theres a button to get a pw by email
[23:44] <michaelni> if you didnt then well i cant help
[23:44] <ubitux> yeah i think i never had one :)
[23:45] <michaelni> you need to talk to mike then
[23:45] <michaelni> thats the disadvantage of it over our wiki
[23:45] <michaelni> delay to get access
[23:47] <ubitux> yep right, just sent an email
[23:49] <michaelni> hey beastd, do you have a wiki multimedia account ? if so you could help improve our SOC page
[23:50] <cone-345> ffmpeg.git 03Senthilnathan M 07master:12a269a5229d: lavu/parseutils: allow MM:SS format for duration in av_parse_time()
[23:51] <cone-345> ffmpeg.git 03Stefano Sabatini 07master:ce84c14ab173: lavu/parseutils: remove spurious "]]" in date format specification
[23:51] <cone-345> ffmpeg.git 03Stefano Sabatini 07master:160def32da43: doc/muxers: fix reference to @anchor defined in another manual
[23:51] <beastd> hi michaelni, i have. will try to have a look at it this weekend.
[23:51] <michaelni> ok thx
[00:00] --- Sat Feb 23 2013