[Ffmpeg-devel-irc] ffmpeg-devel.log.20150506

burek burek021 at gmail.com
Thu May 7 02:05:02 CEST 2015


[00:27:29 CEST] <cone-890> ffmpeg 03Anton Khirnov 07master:3735b5c61677: Revert "lavfi: always check av_expr_parse_and_eval() return value"
[00:27:30 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:af6739d6daf7: Merge commit '3735b5c616770429572f86aabdaec39c6ebb8818'
[01:16:49 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:9e66b39aa87e: diracdec: avoid overflow of bytes*8 in decode_lowdelay
[02:27:06 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07master:d93181ef3eac: diracdec: check if reference could not be allocated
[03:03:48 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07master:a2190de52d6c: avcodec/diracdec: Fix types and wraparounds in dirac_decode_picture_header()
[03:18:04 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:0d3a7dd26490: aasc: return correct buffer size from aasc_decode_frame
[03:18:05 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:97010c74cbff: alsdec: limit avctx->bits_per_raw_sample to 32
[03:18:06 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:4ba11b944e46: Merge commit '0d3a7dd26490156b607541dd2e1faeaa0fc61a88' into release/2.4
[03:18:07 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:1cb470934a26: alsdec: limit avctx->bits_per_raw_sample to 32
[03:18:08 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:8d3a967635c3: Merge commit '97010c74cbff177b58daf9a092b4e37a7da26f85' into release/2.4
[03:33:51 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:7b66cf5ce7fd: aacpsy: correct calculation of minath in psy_3gpp_init
[03:33:52 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:97d7c8d58488: Merge commit '7b66cf5ce7fdb8b3fa13459aab3f4d6ab559f1ea' into release/2.4
[03:48:21 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:41a89cba6086: alsdec: check sample pointer range in revert_channel_correlation
[03:48:22 CEST] <cone-890> ffmpeg 03Andreas Cadhalpun 07release/2.4:378ee3bad5b9: alsdec: only adapt order for positive max_order
[03:48:23 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:843b860ad974: Merge commit '41a89cba6086de2bd24f9ec7e21200fa162505e9' into release/2.4
[03:48:24 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:146648771c95: Merge commit '378ee3bad5b99e8f90864af9bc851590e0f64825' into release/2.4
[04:12:53 CEST] <cone-890> ffmpeg 03Anton Khirnov 07release/2.4:8b86c2ed2eb2: Revert "lavfi: always check av_expr_parse_and_eval() return value"
[04:12:54 CEST] <cone-890> ffmpeg 03Michael Niedermayer 07release/2.4:f28d3f98c5bb: Merge commit '8b86c2ed2eb29dce137b26e66a9bddd50f72817f' into release/2.4
[11:51:21 CEST] <cone-461> ffmpeg 03Dave Yeo 07master:22a0387df29c: OS/2:Makedef.cmd cleanup
[12:30:06 CEST] <Daemon404> i really enjoy when people just ignore reviews that say their patch is malformed and to use send-email
[12:30:11 CEST] <Daemon404> and just paste it in the client every time
[12:30:27 CEST] <JEEBsv> lol
[12:34:10 CEST] <wm4> they probably don't even get what the issue is?
[12:34:36 CEST] <Daemon404> ... i explicitly told them to use send-email
[12:34:39 CEST] <Daemon404> it's a direct instruction
[13:41:20 CEST] <cone-461> ffmpeg 03Nicolas George 07master:7971fa9ce0ea: lavd/xcbgrab: fix comparison with screen size.
[14:08:25 CEST] <kierank> Daemon404: shouldn't you be in berlin now
[15:45:33 CEST] <kierank> j-b: lol kaltura violating gpl
[15:49:00 CEST] <kierank> https://github.com/kaltura/platform-install-packages/issues/392
[15:49:29 CEST] <j-b> kierank: my my my
[15:51:15 CEST] <kierank> just to make it clear they do --enable-nonfree --enable-libfdk-aac  twice in their build
[15:54:55 CEST] <j-b> which makes them not allowed to use ffmpeg anymore
[15:54:57 CEST] <j-b> at all.
[15:57:07 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:890b874889bc: avutil/internal: Suppress empty body warning from emms_c()
[16:39:46 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:0a6b410edf79: avcodec/vp8: Suppress empty body warnings
[16:39:47 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:3c7d3d27be98: tools/cws2fws: Use a similar implementation in dbgprintf() as in ff_dlog()
[17:00:29 CEST] <cone-461> ffmpeg 03Thomas Guillem 07release/2.2:bac0850fbf87: matroskadec: fix crash when parsing invalid mkv
[17:00:30 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:941972d9e980: aasc: return correct buffer size from aasc_decode_frame
[17:00:31 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:b34afd94ec90: Merge commit 'bac0850fbf870d4354d0f68840c0b1c9c1425c4a' into release/2.2
[17:00:32 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:55376ef978f8: Merge commit '941972d9e9806e30f978e8fd033671fd30a80a07' into release/2.2
[17:07:00 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:d0e66cb1c7cf: alsdec: limit avctx->bits_per_raw_sample to 32
[17:07:01 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:b0b92ee04d1e: aacpsy: correct calculation of minath in psy_3gpp_init
[17:07:02 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:645bc997722e: alsdec: limit avctx->bits_per_raw_sample to 32
[17:07:03 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:74409bff4ce9: Merge commit 'd0e66cb1c7cf2a9ef12963cdc6fd49a35599a9dd' into release/2.2
[17:07:04 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:36c5bb6a721a: Merge commit 'b0b92ee04d1ee49a5d8b1aca718ea7d7d44932aa' into release/2.2
[17:26:16 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:d36d3ae02ca7: alsdec: check sample pointer range in revert_channel_correlation
[17:26:17 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:eaa9693fed36: alsdec: only adapt order for positive max_order
[17:26:18 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:f0bf5c538a85: alsdec: ensure channel reordering is reversible
[17:26:19 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:f0cfa5d98a69: avcodec/alsdec: Use av_mallocz_array() for chan_data to ensure the arrays never contain random data
[17:26:20 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:deb0f487bdd7: alsdec: validate time diff index
[17:26:21 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:3a0a2c2586cd: alsdec: check sample pointer range in revert_channel_correlation
[17:26:22 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:0756fa87bffd: Merge commit 'd36d3ae02ca7d535de2b26fa0b3ab44387d19b6a' into release/2.2
[17:26:23 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07release/2.2:e93c46b3ccc1: alsdec: only adapt order for positive max_order
[17:26:24 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07release/2.2:2a947c938261: Merge commit 'eaa9693fed36c29dfbac896946de31eb186ac5d0' into release/2.2
[18:00:49 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07master:4b13a542a27d: dirac: use and forward error codes
[18:03:25 CEST] <ubitux> mmh we can't adjust fps without decoding with ffmpeg?
[18:04:41 CEST] <ubitux> like a basic setpts even without eval, just a factor
[18:06:32 CEST] <ubitux> typically doing something like ffmpeg -r 120 -i 30fps.nut 120fps.nut 
[18:06:39 CEST] <ubitux> but with -c copy
[18:07:08 CEST] <BBB> michaelni: as always, github has a mergeable tree if you want it for the net patchset
[18:07:26 CEST] <BBB> https://github.com/rbultje/ffmpeg/commits/vp9-profile23-wip
[18:08:00 CEST] <BBB> michaelni: and fate samples are at http://downloads.webmproject.org/test_data/libvpx/vp92-2-20-1{0,2}bit-yuv420.webm
[18:08:01 CEST] <BBB> and
[18:08:25 CEST] <BBB> http://downloads.webmproject.org/test_data/libvpx/vp93-2-20-1{0,2}bit-yuv4{22,40,44}.webm
[18:08:34 CEST] <BBB> michaelni: let me know if you need anything else
[18:12:12 CEST] <cone-461> ffmpeg 03Andreas Cadhalpun 07master:75fc81c83185: diracdec: check that block length is valid
[18:12:24 CEST] <j-b> yay, MORE PIX_FMT
[18:12:30 CEST] Action: j-b summons wm4 
[18:12:37 CEST] <BBB> sorry
[18:12:38 CEST] <wm4> huh
[18:12:56 CEST] <BBB> not my choice
[18:13:06 CEST] <wm4> oh dear
[18:13:14 CEST] <wm4> yuv440p10be
[18:13:18 CEST] <wm4> the most useful format ever!
[18:13:19 CEST] <j-b> again, LE vs BE
[18:13:25 CEST] <j-b> Why, why why?
[18:13:28 CEST] <BBB> edontcare
[18:13:32 CEST] <wm4> that's 1/4 subsampling isn't it
[18:13:34 CEST] <BBB> no
[18:13:37 CEST] <wm4> ah no
[18:13:41 CEST] <BBB> its 422 transposed
[18:13:44 CEST] <wm4> +        .log2_chroma_w = 0,
[18:13:44 CEST] <wm4> +        .log2_chroma_h = 1,
[18:13:48 CEST] <wm4> why does this exist?
[18:13:54 CEST] <BBB> I cannot explain
[18:14:04 CEST] <wm4> please whack google for me
[18:14:14 CEST] <BBB> I agree that its a stupid format
[18:14:31 CEST] <BBB> I think some engineer was like, whoa, wouldnt ie be cool if subsampling was a variable
[18:14:31 CEST] <j-b> Can't we trivially convert to 422 ?
[18:14:36 CEST] <wm4> didn't read the patches, but I suppose the format lets you specify arbitrary depths/subsampling
[18:14:43 CEST] <wm4> so we need to add all combinations thereof?
[18:14:46 CEST] <BBB> and then he created 444, 422 (yay!) and happened to create 440 also (oh well)
[18:14:49 CEST] <BBB> and then he moved on
[18:14:59 CEST] <BBB> right
[18:15:09 CEST] <BBB> there is no useful use case for 440 that I can think of
[18:15:13 CEST] <j-b> Do we _really_ need both BE and LE of this stupidity?
[18:15:20 CEST] <BBB> but theres conformance suite samples so Im going to assume that its required to be supported
[18:15:31 CEST] <BBB> BE/LE is cargo cult, feel free to fix it but outside scope of this patch
[18:16:03 CEST] <wm4> j-b: many encoders/decoders for stupid legacy image formats need them
[18:16:17 CEST] <wm4> so cleaning up this shit would take a while
[18:17:28 CEST] <j-b> wm4: of course, but why add it for new ones?
[18:17:30 CEST] <j-b> like 440
[18:17:31 CEST] <BBB> Im also wondering if I should consolidate some of the dsp functions
[18:18:07 CEST] <BBB> like copy for 8bpp and 16bpp is identical if size is double
[18:18:13 CEST] <wm4> j-b: because then you'd get swapped endian on a BE machine
[18:18:17 CEST] <BBB> and intra pred for 10 and 12bpp are identical except for tm
[18:18:27 CEST] <wm4> and we all care about BE machines, right
[18:18:29 CEST] <j-b> wm4: which BE machine?
[18:19:09 CEST] <j-b> wm4: and, so? BE machines decoding 440 10bits, would get a swap to do...
[18:19:32 CEST] <BBB> the format is always native
[18:19:41 CEST] <wm4> then the format would be native endian, which would probably conflict with raw encoders or some shit
[18:19:43 CEST] <BBB> the only case where Be/LE comes in is if we store it in a raw .yuv file
[18:19:51 CEST] <BBB> and read that on a machine of different endianness
[18:20:09 CEST] <BBB> in any other case the machine only ever sees the native form
[18:20:17 CEST] <wm4> currently, a pixfmt is supposed to define a byte representation, endian-independent
[18:20:37 CEST] <wm4> (not that I agree with this, but that's how it is)
[18:20:59 CEST] <wm4> Libav is open to cleaning it up
[18:22:56 CEST] <BBB> I encourage you guys to clean it up
[18:23:00 CEST] <BBB> but its orthogonal to the patchset
[18:23:25 CEST] <BBB> its like bringing up ambiguous cosmetic shit in new decoder submissions
[18:24:04 CEST] <j-b> I kind of disagree
[18:24:05 CEST] <BBB> please align horizontally and shit like that, Im completely allergic to that sort of stuff
[18:24:38 CEST] <j-b> Your patch is adding a new pixfmt to libavutil, it's not orthogonal to the patchset
[18:24:50 CEST] <j-b> and it's very very different from stupid cosmetic discussions
[18:25:11 CEST] <j-b> Once it's in the headers, you're screwed
[18:25:28 CEST] <BBB> yuv420p9be is already in the headers
[18:25:31 CEST] <BBB> so the problem exists
[18:25:37 CEST] <BBB> if you want to fix it, fix it for all of them
[18:25:45 CEST] <BBB> saying that 440p10le is somehow special, is silly
[18:26:46 CEST] <wm4> yeah, in theory the addition of your pixfmt is fine
[18:26:54 CEST] <wm4> it follows the existing rules
[18:27:21 CEST] <BBB> I will whack google about the more general concept of adding 440
[18:27:31 CEST] <BBB> I agree that 440 is stupid and unnecessary
[18:28:14 CEST] <wm4> they are probably fascinated by the idea that they can implement different subsampling so generically and easily
[18:28:25 CEST] <wm4> but that doesn't mean it's a good idea to allow such obscure/useless formats
[18:30:49 CEST] <iive> BBB: is there point in using something else than native endiannes for decoder?
[18:31:43 CEST] <iive> aka, are bitstreams different for 420p10be and 420p10le ?
[18:31:54 CEST] <BBB> decoders only output native
[18:32:05 CEST] <BBB> but rawvideo from a yuv file could be anything
[18:32:53 CEST] <iive> i'm not against adding new pixfmts, rgb vs bgr is also endian issue.
[18:33:19 CEST] <wm4> iive: we've discussed why both are needed above
[18:35:51 CEST] <BBB> wm4: https://code.google.com/p/webm/issues/detail?id=996
[18:37:34 CEST] <wm4> huh, they're still on google code?
[18:38:37 CEST] <wm4> also, it's not strictly about the needed amounts of code, but the explosion of (obscure) pixel formats, which directly burdens applications
[18:39:14 CEST] <wm4> even vlc is complaining!
[18:39:50 CEST] <BBB> I dont think it matters what its about
[18:39:51 CEST] <BBB> it should go
[18:40:10 CEST] <BBB> feel free to voice your outrage in that bug report :)
[18:40:38 CEST] <wm4> you pasted the rage into your bug report
[18:40:46 CEST] <BBB> more rage is good
[18:42:28 CEST] <michaelni> BBB, the patchset needs some version check i think, it fails building libavcodec/libvpxdec.c here with libavcodec/libvpxdec.c:75:10: error: VPX_IMG_FMT_I440 undeclared (first use in this function)
[18:44:04 CEST] <BBB> hm...
[18:44:08 CEST] <BBB> thats odd
[18:44:16 CEST] <BBB> let me check
[18:45:39 CEST] <BBB> youre right, it was added in libvpx commit a30774c50d2b1b626b05a99dcd29aa766f0b223b
[18:45:54 CEST] <BBB> what do you want me to do? update configure check?
[18:48:52 CEST] <Daemon404> kierank, no
[18:48:54 CEST] <Daemon404> im flying tomorrow
[18:48:57 CEST] <Daemon404> im not going to the first day
[18:49:07 CEST] <kierank> lol
[18:49:07 CEST] <Daemon404> why would i want to sit in the middle of a circlejerk for longer than needed
[18:49:09 CEST] <michaelni> BBB, no idea, whatever you prefer, or ask James Zern (he is listed in MAINTAINERs)
[18:49:19 CEST] <wm4> is this about the smpte conference?
[18:49:24 CEST] <kierank> Daemon404: because v-nova
[18:49:26 CEST] <kierank> dugh
[18:49:33 CEST] <kierank> changes everything
[18:49:33 CEST] <michaelni> BBB, also you need a update for fate-filter-pixfmts-crop and a few others
[18:49:45 CEST] <BBB> oh right
[18:49:48 CEST] <BBB> I forgot
[18:50:50 CEST] <Daemon404> kierank, they even offered me a job
[18:50:52 CEST] <Daemon404> lul
[18:52:23 CEST] <j-b> Daemon404: haha. you accepted, yes?
[18:52:33 CEST] <kierank> amazing how much noise you can make over vapourware
[18:52:44 CEST] <Daemon404> j-b, lol
[18:53:12 CEST] <Daemon404> i should probab;ly practice my slides at some point
[18:53:14 CEST] <Daemon404> probably.
[18:53:23 CEST] <j-b> Daemon404: you should have met them
[18:53:29 CEST] <j-b> and report the bullshitware
[18:53:36 CEST] <Daemon404> vittorio met them iirc
[18:54:04 CEST] <j-b> ah?
[18:54:20 CEST] <Daemon404> id rather not ;)
[18:54:27 CEST] <kierank> i saw a lot of them at tv connect
[18:54:43 CEST] <j-b> they are on the awarding list BS
[18:55:06 CEST] <kierank> you just pay money and you get an award
[18:55:12 CEST] <kierank> it's not hard
[18:55:16 CEST] <kierank> they call me all the time
[18:55:34 CEST] <Daemon404> sounds like a posh special ed school
[20:22:28 CEST] <cone-461> ffmpeg 03Nicolas George 07master:01fdfa51aca9: xcbgrab: Accept geometries matching the screen size
[20:22:29 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:958ddef46c22: Merge commit '01fdfa51aca9086e04bd354fe3f103a49352c085'
[20:31:28 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:d00f1e0fc10b: quickdraw: Convert to bytestream2
[20:31:29 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:d423fc12e6e1: Merge commit 'd00f1e0fc10bb945cdf140a6c320eb0c9cd25162'
[20:50:39 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:34efb8a169e4: quickdraw: Support direct pixel blocks
[20:50:40 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:39e89f521159: Merge commit '34efb8a169e4551326e069be47125c6c2cb7ab90'
[20:51:18 CEST] <jamrial> BBB: yeah, works now
[20:58:56 CEST] <kierank> j-b: asking the lawyers: https://github.com/kaltura/platform-install-packages/issues/392#issuecomment-99524232
[21:00:20 CEST] <BBB> jamrial: do you know anything about configure version checks to get libvpxs 440 symbol to be there?
[21:03:29 CEST] <jamrial> not really. VPX_IMG_FMT_HIGHBITDEPTH is currently checked for profile 2, so it probably guarantees profile 1 440 but not profile 3
[21:05:34 CEST] <jamrial> and the check should be done in libvpx*.c rather than configure, where we just check for a new enough version to enable the vp8/vp9 decoders and encoders
[21:08:33 CEST] <BBB> the same commit introduced highbitdepth 440 as well as normal 440
[21:08:43 CEST] <BBB> so I think the define was added after high bitdepth in the code
[21:08:52 CEST] <BBB> also, running all of fate is REALLY slow
[21:08:58 CEST] <BBB> as in, we have a shitton of tests
[21:09:02 CEST] <BBB> are all of these tests useful?
[21:11:02 CEST] <BBB> michaelni: pushed new github tree which adds/updates relevant references for 440
[21:16:18 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:7b734ee55dbb: lavf: Open PICT images with Quickdraw
[21:16:19 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:ffed987a4d4a: Merge commit '7b734ee55dbb8476d7ad63c7daf55c534cf82d5d'
[21:29:34 CEST] <jamrial> BBB: checking for VPX_IMAGE_ABI_VERSION >= 3 should work
[21:30:44 CEST] <BBB> ok, pushed a new version (to github) that does that
[21:30:54 CEST] <jamrial> that define value was bumped about two months after 440 support was added. not ideal, but i guess it's better than nothing
[21:31:04 CEST] <BBB> its good enough I guess
[21:46:07 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:d81fb63d8769: fate: Add a PICT test
[21:46:08 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:20ce69e5f85d: Merge commit 'd81fb63d87692765c004c19934b49427df434a07'
[22:28:57 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:04070dbca068: libx265: Fix 'braces around scalar initializer' warning
[22:28:58 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:1caff5731d9a: Merge commit '04070dbca0688ab1e24528ce5c135254a9a79c47'
[22:35:48 CEST] <cone-461> ffmpeg 03Vittorio Giovara 07master:6c8c7c246280: dump: Print AVAudioServiceType side data
[22:35:49 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:0318013afcff: Merge commit '6c8c7c246280cdcd23e4d99006b95245ae837278'
[22:44:26 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:e0010bb91fc0: mux: Do not leave stale side data pointers in ff_interleave_add_packet()
[22:44:27 CEST] <cone-461> ffmpeg 03Michael Niedermayer 07master:2284c38e4c68: Merge commit 'e0010bb91fc0b067be8aefd2f77e6838537da433'
[00:00:00 CEST] --- Thu May  7 2015


More information about the Ffmpeg-devel-irc mailing list