[Ffmpeg-devel-irc] ffmpeg-devel.log.20140309

burek burek021 at gmail.com
Mon Mar 10 02:05:02 CET 2014


[00:57] <cone-459> ffmpeg.git 03Diego Biurrun 07master:a80145342342: g2meet: K&R formatting cosmetics
[00:57] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:17a5dbacf50c: Merge commit 'a801453423420915425f902bf9f9cbe982a55d87'
[01:04] <cone-459> ffmpeg.git 03Keiji Costantini 07master:b4d372e091f6: rv10: Forward error from rv10_decode_packet
[01:04] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:0c1353ae338b: Merge commit 'b4d372e091f6b30758db2a43a5a9fe2510ec2b13'
[01:15] <cone-459> ffmpeg.git 03Keiji Costantini 07master:8c09532672d7: rv10: K&R formatting cosmetics
[01:15] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:2beb81522e59: Merge commit '8c09532672d7d3196ec52afefa2c96633bc17b62'
[01:23] <cone-459> ffmpeg.git 03Janne Grunau 07master:61985ad72c47: arm: hpeldsp: fix put_pixels8_y2_{,no_rnd_}armv6
[01:23] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:b39e89502437: Merge commit '61985ad72c47bbb668f2d3923bf5c9df83e79323'
[01:50] <cone-459> ffmpeg.git 03John Stebbins 07master:0092c1dd8dac: matroskaenc: allow override of "writing application" tag
[01:50] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:4cb6814160d0: Merge commit '0092c1dd8dac2d9e185b58503b447a0d3fb5230d'
[02:03] <cone-459> ffmpeg.git 03John Stebbins 07master:565e0c6d866c: movenc: allow override of "writing application" tag
[02:03] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:9827f3db88c0: Merge commit '565e0c6d866ce08d4b06427456d3d1f4fd856e9c'
[02:16] <cone-459> ffmpeg.git 03Pierre Lejeune 07master:0e0cefb22216: build: Use pkg-config for openjpeg
[02:16] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:9c99bc1f3fb9: Merge remote-tracking branch 'qatar/master'
[03:26] <cone-459> ffmpeg.git 03Marton Balint 07master:ae017c26325f: libzvbi-teletextdec: use av_dlog where possible
[03:26] <cone-459> ffmpeg.git 03Marton Balint 07master:085ca7dcdbf9: libzvbi-teletextdec: split dvb packet to slices
[03:26] <cone-459> ffmpeg.git 03Michael Niedermayer 07master:ea9399f8f7dc: Merge remote-tracking branch 'cus/stable'
[12:57] <ubitux> wow, arm can be really awful to read sometimes when you're not used to read it
[12:57] <ubitux> like, vld1.32 {d20[]}, [r0,:32], r2
[12:58] <ubitux> is this some kind of perl superset?
[13:01] <wbs> * can be awful to read when you're not used to read it, news at 11
[13:02] <nevcairiel> it does look quite a bit more weird then x86
[13:02] <wbs> it has a different syntax yes - once you know what it does it's way more simple and structured to deal with
[13:02] <wbs> ... afaik
[13:03] <ubitux> wbs: the point was that intel for instance was more readable even when not used to read; i'm not really seriously ranting against it :p
[13:04] <ubitux> not until i understand it :p
[13:27] <cone-111> ffmpeg.git 03Janne Grunau 07master:4506a854a4d8: arm: vp3: remove incorrect const in ff_vp3_idct_dc_add_neon declaration
[13:27] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:fc1d7811ef4a: Merge commit '4506a854a4d846692ba71daeeff661dc214c8fa2'
[13:31] <BBB> ubitux: I agree actually - I think the main problem is that it supports some extra things like inline shifting of GPR values when using them (similar to the mov m1, [x*8] in x86) or pre/post-increment of GPRs, and some more odd stuff, and its notation for that is "different" and you have to get used to it
[13:31] <BBB> the inline shifting isn't so weird in notation
[13:31] <BBB> pre/post increment is just pure garbage if you're not used to it
[13:37] <cone-111> ffmpeg.git 03Janne Grunau 07master:dfe224f377be: aarch64: get_cabac inline asm
[13:37] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:cbee7ff0b7df: Merge commit 'dfe224f377be3e45758c69d881ca7874b82d647a'
[14:07] <cone-111> ffmpeg.git 03Luca Barbato 07release/1.2:78197501572a: avidec: Let the inner dv demuxer take care of discarding
[14:26] <cone-111> ffmpeg.git 03Janne Grunau 07master:634d9d8b3989: arm: get_cabac inline asm
[14:27] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:fa4f57399774: Merge commit '634d9d8b398982647b3d7160641198744901d8d8'
[14:42] <wm4> seeking in tak seems pretty unreliable...
[14:44] <nevcairiel> wasnt tak one of those formats with an optional seeking tabel
[14:44] <nevcairiel> i might be confusing it
[14:45] <nevcairiel> guess i was
[14:45] <wm4> if it does, lavf doesn't use it
[14:45] <nevcairiel> otherwise its just a vbr raw format, there is no good way to seek in those
[14:45] <wm4> hm, bad
[14:45] <wm4> is the format still closed?
[14:46] <nevcairiel> I have been wondering if its worth implementing a memory cache with indexing for "small" audio files, small being like everything up to 100mb or something, so that you can seek in those at least
[14:46] <nevcairiel> people always complain about vbr mp3 or vbr aac to me as well
[14:47] <wm4> pretty much depends whether the file is on fast storage or not
[14:47] <wm4> but if so, indexing "small" files would probably be pretty quick
[14:51] <wm4> seems like the closed source tak decoder seems to claim it can seek sample-exact
[14:52] <wm4> other than that there's a shitty SDK and a pascal source file which apparently documents the bitstream or parts of it
[14:55] <wm4> there was a seek table, but it was removed later, and it appears now you're supposed to crawl the raw stream yourself, searching for a header (which contains a frame number and a crc)
[14:56] <wm4> and that's where I stop caring
[14:59] <wm4> nevcairiel: maybe creating an index in advance would be a nice optional feature for libavformat
[14:59] <wm4> it could use the existing generic index stuff (although that looks yucky to mess with)
[15:23] <cone-111> ffmpeg.git 03Janne Grunau 07master:c708b5403346: timer: use mach_absolute_time as high resolution clock on darwin
[15:23] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:c2923100f5a1: Merge commit 'c708b5403346255ea5adc776645616cc7c61f078'
[15:23] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:4159f702a7b8: avutil/timer: Fix units for x86 after c708b5403346255ea5adc776645616cc7c61f078
[15:38] <cone-111> ffmpeg.git 03Janne Grunau 07master:74cc90190574: float_dsp: fix errors in documentation
[15:38] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:8dfd0e62f5d3: Merge remote-tracking branch 'qatar/master'
[16:39] <BBB> nevcairiel: I think that's a good idea, it could be filled in the bg while the fg thread is already serving data, and could then be format-agnostic
[16:39] <BBB> nevcairiel: just needs someone doing it
[21:08] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:920c01adce6c: hevc: Use get_bits_long() in decode_vui()
[21:08] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:e15a57b67a26: Merge commit '920c01adce6c273fc043513ff237a6266e612152'
[21:14] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:6cc94e971933: hevc: use av_mallocz() for allocating tab_ipm
[21:14] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:26568c04a827: Merge commit '6cc94e971933cd38c452172bb048bf760e65cc3e'
[21:41] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:ed06e5d92b4c: hevc: Do not turn 32bit timebases into negative numbers
[21:41] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:06219db9351c: Merge commit 'ed06e5d92b4c67b49068d538461fbbe0a53a8c5e'
[21:53] <ubitux> i've a hard time following the arm doc; what's the diff between "[r1,:128]" and "[r1,128]"?
[21:54] <ubitux> i see "[r1,off]" and "[r1:align]"
[21:54] <ubitux> what's this ",:"? it means "aligned [r1,off]" or something?
[22:06] <cone-111> ffmpeg.git 03Vittorio Giovara 07master:4d33873c2990: hevc: make pps/sps ids unsigned where necessary
[22:06] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:eaaa5801efec: Merge commit '4d33873c2990b8d6096f60fef384f0efc4482b55'
[22:07] <cone-111> ffmpeg.git 03Ben Littler 07master:c09804684a47: isom: add xd51 hdcam
[22:07] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:8a5397d88b8c: Merge commit 'c09804684a47e6572ee82ded5d84c8b282815da6'
[22:12] <cone-111> ffmpeg.git 03Carl Eugen Hoyos 07master:eeae858596a5: isom: Support Sony XAVC in mov
[22:12] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:2f41b186963d: Merge commit 'eeae858596a5ce5a84797283f590a16add5704f4'
[22:15] <cone-111> ffmpeg.git 03Piotr Bandurski 07master:650901239843: isom: add Radius DV YUV FourCCs
[22:15] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:6783b3eb761b: avformat/isom: remove duplicated line
[22:15] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:2030b3a4abd8: Merge commit '6509012398435252979e149ea4a73438d2107600'
[22:20] <cone-111> ffmpeg.git 03Piotr Bandurski 07master:31e558fbcfec: isom: add "NO16" FourCC
[22:20] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:d32245e5b12c: Merge commit '31e558fbcfec8fa9fddf564ab8360384706e41c0'
[22:22] <cone-111> ffmpeg.git 03Carl Eugen Hoyos 07master:073cd30fb4dc: isom: Support MPEG-2 video mov files from FPC5
[22:22] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:76367f453f6f: Merge commit '073cd30fb4dc1c2a885c62540e3cfaac1f17c21b'
[22:23] <cone-111> ffmpeg.git 03Piotr Bandurski 07master:db32f14c3dfd: isom: add "MNG" FourCC
[22:23] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:c68e53f27281: Merge commit 'db32f14c3dfd75290b57a086183d6577457beac4'
[22:24] <cone-111> ffmpeg.git 03Mark Himsley 07master:360022bd3b89: isom: lpcm in mov default to big endian
[22:24] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:9f71e5d39e0e: Merge commit '360022bd3b894cc01ea112b275fa4c8f53881808'
[22:26] <cone-111> ffmpeg.git 03Piotr Bandurski 07master:8a0bc40545a2: libspeex: support ZygoAudio (quality 10 mode)
[22:26] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:858d9947e4fe: Merge commit '8a0bc40545a216a49f43db5d9a42a63961333354'
[22:28] <cone-111> ffmpeg.git 03Baptiste Coudurier 07master:9e71cc81f365: movdec: handle 0x7fff langcode as macintosh per the specs
[22:28] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:0cf6b3b5b79f: Merge commit '9e71cc81f3655cacf0f91860fba3043f13b64059'
[22:30] <cone-111> ffmpeg.git 03Ben Littler 07master:fb0a988f94c3: riff: IPJ2 decodes with JPEG2000 decoder
[22:30] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:4aa48e467b1c: Merge commit 'fb0a988f94c3b8907610172295a7f6bf78c57105'
[22:31] <cone-111> ffmpeg.git 03James Zern 07master:3e0175d89a6c: riff: add VP9 fourcc
[22:31] <cone-111> ffmpeg.git 03Michael Niedermayer 07master:7a69ffaf0df4: Merge commit '3e0175d89a6c5ee8393ae05e1b40abdca1204919'
[22:37] <cone-29> ffmpeg.git 03Tim Walker 07master:1d9014f0b008: movenc: use 'hev1' tag for HEVC in MODE_MOV.
[22:37] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:ce1ad8d775ba: Merge commit '1d9014f0b008485eac4c19d5f5e11ede59237167'
[23:06] <cone-29> ffmpeg.git 03Tim Walker 07master:20b40a597cdd: movenc: write hvcC tag for HEVC.
[23:06] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:1b1d77ae9b38: Merge commit '20b40a597cdd4969cf1147d7c7efee2b6232524b'
[23:06] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:cb403b257038: avformat: fix hevcs use of golomb from avformat
[23:15] <ubitux> oh, mail from qatar on ffmpeg-user.
[23:16] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-user/2014-March/020425.html
[23:17] <ubitux> (wtf)
[23:18] <JEEBsv> lol
[23:18] <Mavrik> Oracle DBA.
[23:19] <ubitux> it looks like a message constructed with markov chains based on indian and chinese mails from ffmpeg-user
[23:29] <BBB> ubitux: I believe [r1, :128] is post-increment
[23:29] <BBB> ubitux: and [r1, 128] is pre-increment
[23:30] <BBB> ubitux: what that means is (in x86 notation) load [r1], then add r1, 128 (post-increment) or add r1, 128 and then load [r1]
[23:30] <BBB> (pre-increment)
[23:30] <ubitux> isn't this done respectively with [r1,off]! and [r1],off?
[23:30] <BBB> hm maybe ! was post... argh
[23:30] <ubitux> ! is pre
[23:30] <BBB> lol
[23:31] <BBB> ok I'm an idiot
[23:31] <ubitux> [r1],off is post
[23:31] <BBB> don't listen to me obviousl
[23:31] <ubitux> but i don't understand how that works with that "align" parameter (the : thing)
[23:32] <ubitux> so far, i understood that: "[R,off]" ’ "R[off]", "[R,off]!" ’ "r+=off; [R]", "[R],off" ’ "R[off]; r+=off"
[23:33] <ubitux> A ’ B with A asm syntax and B pseudo code how i understand it
[23:33] <cone-29> ffmpeg.git 03Tim Walker 07master:b6c61fb83e87: movenc: enable Annex B to MP4 conversion for HEVC tracks.
[23:34] <cone-29> ffmpeg.git 03Tim Walker 07master:4f3db5d3418a: movenc: allow muxing HEVC in MODE_MP4.
[23:34] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:a379813ceed6: Merge commit 'b6c61fb83e876d404ac3b0b3657ebfcafdcd1926'
[23:34] <ubitux> it also seems you can have mixes like "[R+off1],off2", which i guess is going to do something like "R[off1+off2]; R+=off1"
[23:34] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:5d5e2bd86242: avformat/hevc: Make return codes consistent and more flexible
[23:34] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:9563e67e0759: Merge commit '4f3db5d3418a24f3b90422e98ad75388052c4284'
[23:34] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:88c8e4afeaf7: avformat/hevc: fix mix of av_malloc() with free()
[23:34] <ubitux> but i'm trying to get that weird ",:" thing
[23:41] <cone-29> ffmpeg.git 03Tim Walker 07master:a823d0948683: matroskaenc: write private data in hvcC format for HEVC.
[23:41] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:38aee6408a67: Merge commit 'a823d0948683bd97dd58556b5740e434166209a8'
[23:42] <BBB> ubitux: I have no idea; maybe ask wbs?
[23:42] <ubitux> i hope to figure out what "vld1.32 {d0[]}, [r0,:32], r2" means by the end of the next week
[23:43] <BBB> http://people.xiph.org/~tterribe/daala/neon_tutorial.pdf
[23:43] <BBB> is quite useful
[23:43] <BBB> (note that in their syntax, : is @)
[23:43] <ubitux> ooh
[23:43] <ubitux> i was trying to learn with the official man
[23:43] <ubitux> but i find it extremely confusing
[23:43] <ubitux> thanks for the link
[23:43] <BBB> specifically slide 19
[23:44] <BBB> vld2.8 {d0, d1} [r0] load 2 d registers stride 2 8 bits from [r0]
[23:45] <ubitux> yeah that part is ok
[23:45] <BBB> so vld1.32 {d0[]} [r0,:32], r2 would probably be something like load consecutive units of 32 bits from r0 aligned by 32 bits (?) into d0
[23:45] <BBB> right, but then the r2...
[23:45] <BBB> post-increment?
[23:46] <BBB> so increment r0 by r2 after the load?
[23:46] <ubitux> yes it's post increment of r2
[23:46] <ubitux> but note the ",:"
[23:46] <ubitux> the ',' is for the offseting
[23:46] <ubitux> and ':' is for the align
[23:50] <BBB> wbs: ? ^^
[23:54] <cone-29> ffmpeg.git 03Tim Walker 07master:558b20d729bc: matroskaenc: enable Annex B to MP4 conversion for HEVC tracks.
[23:54] <cone-29> ffmpeg.git 03Michael Niedermayer 07master:fb0023639ea5: Merge remote-tracking branch 'qatar/master'
[00:00] --- Mon Mar 10 2014


More information about the Ffmpeg-devel-irc mailing list