[Ffmpeg-devel-irc] ffmpeg-devel.log.20151108

[burek]

[00:08:11 CET] Action: BBB wonders what ubitux is working on
[00:12:04 CET] <Timothy_Gu> kierank: do you have a test case for https://trac.ffmpeg.org/ticket/1869 ?
[00:12:40 CET] <Timothy_Gu> like a yuv raw file
[00:12:41 CET] <kierank> https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket1869/
[00:12:47 CET] <kierank> oh
[00:13:19 CET] <Timothy_Gu> ok thx
[00:16:18 CET] <kierank> I have some yuv10 samples somewhere as well
[00:16:50 CET] <kierank> yuv10 == v210x
[00:17:23 CET] <Timothy_Gu> kierank: for that ticket, do you think I should create a new demuxer for v210?
[00:17:33 CET] <Timothy_Gu> also what's a parser?
[00:17:35 CET] <kierank> yes a simple one that calculates the frame size
[00:17:48 CET] <kierank> a parser is there to separate a bunch of data into different frames
[00:18:12 CET] <kierank> so I just have opus data that comes from a ts demux but I have no idea where frames begin or end 
[00:18:38 CET] <Timothy_Gu> do you mean there's no concept of "frame" in ts?
[00:18:48 CET] <Timothy_Gu> or can a frame be spread into several packets
[00:18:54 CET] <kierank> not really related to ts but it's for any kind of data stream you get
[00:18:55 CET] <kierank> yes
[00:18:59 CET] <Timothy_Gu> ah
[00:19:00 CET] <kierank> there are others like raw ac3
[00:19:33 CET] <Timothy_Gu> should I/can I write a probe function for v210?
[00:19:40 CET] <Timothy_Gu> I can't right?
[00:20:21 CET] <kierank> maybe from the file extension .v210
[00:20:27 CET] <kierank> and .yuv10 for v210x
[00:20:29 CET] <kierank> but that's it
[00:21:34 CET] <Timothy_Gu> and you mentioned something about endian issue for v210x and v210
[00:21:38 CET] <Timothy_Gu> what is it exactly?
[00:22:27 CET] <kierank> afaik v210 and v210x are similar but with endians reversed
[00:22:44 CET] <Timothy_Gu> so
[00:23:03 CET] <Timothy_Gu> v210: 012345678901234567890123456789xx
[00:23:27 CET] <Timothy_Gu> v210x:xx98765432109876543219876543210
[00:23:38 CET] <Timothy_Gu> v210x:xx987654321098765432109876543210
[00:23:41 CET] <Timothy_Gu> ?
[00:25:35 CET] <kierank> so if that was a bit pattern in increasing memory order, I believe yes
[00:26:21 CET] <Timothy_Gu> ok
[00:26:56 CET] <Timothy_Gu> I need more practice in reading bit manipulation code
[00:27:28 CET] <kierank> v210 and v210x are confusing because for historical reasons v210x decodes to YUV422P16
[00:28:14 CET] <kierank> http://dsplab.diei.unipg.it/software/pyuv_raw_video_sequence_player
[00:28:18 CET] <kierank> that's the viewer I use for 10-bit video
[00:33:37 CET] <Timothy_Gu> what should I set for time_base?
[00:33:46 CET] <Timothy_Gu> if I should set anything
[00:34:02 CET] <Timothy_Gu> or does avpriv_set_pts_info automatically do it?
[00:34:28 CET] <kierank> shouldn't set anything I think
[00:34:37 CET] <kierank> I would guess ffmpeg.c would give it a default timebase
[00:35:08 CET] <Timothy_Gu> I'm basically working on a copy of rawvideodec.c
[00:35:19 CET] <Timothy_Gu> and it has avpriv_set_pts_info(st, 64, s->framerate.den, s->framerate.num);
[00:35:30 CET] <Timothy_Gu> s->framerate can be sset by a user
[00:35:53 CET] <Timothy_Gu> and reading the source of avpriv_set_pts_info seems like it sets the time base according to the framerate
[00:36:00 CET] <kierank> yes
[00:36:19 CET] <Timothy_Gu> ok
[00:55:48 CET] <Timothy_Gu> kierank: so, I got this working: ffmpeg -s 1920x1080 -f v210 -i 1920x1080_ducks_v210.yuv -f null -
[00:55:57 CET] <Timothy_Gu> it outputs 10 frames now instead of 6
[00:56:03 CET] <Timothy_Gu> so am I done?
[00:56:09 CET] <kierank> I suspect yes
[03:47:11 CET] <cone-977> ffmpeg 03Timothy Gu 07master:3d20f8e7c0e2: Add pixblockdsp checkasm tests
[11:03:38 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:a66b243d5223: avutil/softfloat: FLOAT_0 should use MIN_EXP
[11:32:58 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:dcf1cf5d24c8: avcodec/aacsbr: Use FLOAT_0
[13:42:12 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:107db5abf321: avutil/softfloat: Correctly set the exponent for 0.0 in av_sqrt_sf()
[13:42:13 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:0269fb11e3de: avutil/softfloat: Check for MIN_EXP in av_sqrt_sf()
[14:29:27 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:4b6ad23609f4: Revert "avutil/softfloat: Check for MIN_EXP in av_sqrt_sf()"
[14:29:28 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:4135a2bfd6d0: avutil/softfloat: Fix typo in av_mul_sf() doxy
[14:29:30 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:a1e3303fc01b: avutil/softfloat: Fix exponent underflow in av_mul_sf()
[14:29:31 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:046218b212a0: avutil/softfloat: Fix exponent underflow in av_div_sf()
[14:29:32 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:596dfe7d6c6b: avutil/softfloat: Add tests for exponent underflows
[14:58:27 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:df2a2117d280: avutil/softfloat: Add test for av_cmp_sf()
[14:58:28 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:cee3c9d29ace: avutil/softfloat: Fix overflows in shifts in av_cmp_sf() and av_gt_sf()
[14:58:28 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:ecfb076141d0: avutil/softfloat: Extend the av_cmp_sf() test to cover a wider range of exponents
[15:17:57 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:791ea23e57f1: avutil/softfloat: Add test for av_gt_sf()
[15:17:58 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:05b05a7a841c: avutil/softfloat: Fix av_gt_sf() with large exponents try #2
[15:17:58 CET] <cone-630> ffmpeg 03Michael Niedermayer 07master:955cdc43a35a: avutil/softfloat: Include negative numbers in cmp/gt tests
[17:59:37 CET] <Daemon404> i always wonder about those "my player cant play this" bgs
[17:59:38 CET] <Daemon404> bugs
[18:54:43 CET] <cone-190> ffmpeg 03Kyle Swanson 07master:6a11c7f1605d: avfilter: add anoisesrc
[21:10:12 CET] <cone-190> ffmpeg 03Andreas Cadhalpun 07master:f3866a14c3c2: softfloat: assert when the argument of av_sqrt_sf is negative
[21:10:13 CET] <cone-190> ffmpeg 03Andreas Cadhalpun 07master:9ac61e73d084: softfloat: handle INT_MIN correctly in av_int2sf
[21:24:18 CET] <durandal_1707> heh, not comitting patch that got approved even if its security related
[21:31:57 CET] <jamrial_> [PATCH 001/264]
[21:31:58 CET] <jamrial_> o_o
[21:33:16 CET] <Daemon404> you clearly dont remember TEP1
[21:34:03 CET] <rcombs> what was TEP1 about
[21:34:14 CET] <Daemon404> adding reference counting
[21:34:22 CET] <rcombs> ah
[21:36:36 CET] <BtbN> On which ML? oO
[21:36:59 CET] <jamrial_> libav-devel
[21:37:26 CET] <rcombs> git hype
[23:02:11 CET] <J_Darnley> Is TEP supposed to be an abbreviation of something?
[23:03:05 CET] <Gramner> The Evil Plan
[23:03:50 CET] Action: J_Darnley wonders if that is sincere or sarcastic
[23:04:41 CET] <kierank> there's a lot of malloc going on now
[23:08:19 CET] <rcombs> "evil" in that it breaks application code
[23:20:06 CET] <Daemon404> kierank, he has to make up for all the malloc he removed with tep1 perhaps
[23:29:16 CET] <pross> oh yeah, patch 1/264
[23:29:17 CET] <durandal_1707> that's insane (mallocs)
[23:31:12 CET] <nevcairiel> 1/264 doesnt add new mallocs
[23:31:25 CET] <nevcairiel> in fact, it adds a function to add sidedata without a malloc
[23:42:32 CET] <pross> nevcairiel: converting chucks of avcodeccontext into semantically grouped sidedata structs will result in more alloc/reallocs.
[23:48:47 CET] <nevcairiel> But that's not 1/264, that patch in particular adds a new function for potentially less mallocs :p
[23:53:14 CET] <kierank> nevcairiel: i thought all side data rquired a malloc?
[23:59:09 CET] <nevcairiel> No one reads apparently
[00:00:00 CET] --- Mon Nov  9 2015