[Ffmpeg-devel-irc] ffmpeg-devel.log.20180202

[burek]

[00:20:37 CET] <cone-276> ffmpeg 03Michael Niedermayer 07master:0a2560a9775b: avcodec/exr: Fix memleaks in decode_header()
[00:20:38 CET] <cone-276> ffmpeg 03Michael Niedermayer 07master:b1bef755f617: avcodec/aacsbr_fixed: Fix overflows in rounding in sbr_hf_assemble()
[00:20:39 CET] <cone-276> ffmpeg 03Xiaohan Wang 07master:caaa40d2c67b: configure: Remove carriage return ('\r') in Windows CC_IDENT
[01:29:19 CET] <mypopydev> @jkqxz Hi
[01:30:03 CET] <mypopydev> Can you help to review https://patchwork.ffmpeg.org/patch/7439/, tks
[08:25:39 CET] <blackelement> Does the rawvideo format contain anything other than standard YUV420P in the output file? I'm seeing usual data and then things get weird pretty quick
[08:28:04 CET] <blackelement> Specifically 3330 and 0d0a hex repeating 
[09:54:22 CET] <kierank> It seems Chrome Embedded Framework lets you build ffmpeg and strip the licence flags
[10:13:46 CET] <atomnuker> license flags?
[10:14:09 CET] <kierank> the configure string and the string telling you what FFmpeg's licence is
[10:22:59 CET] <atomnuker> nasty, but I guess they needed the space
[10:23:25 CET] <atomnuker> you can still just look up the symbols and figure out if it has any GPL ones
[10:25:56 CET] <atomnuker> or if they strip off everything you can still try to find the flac x86 asm instructions
[10:59:30 CET] <kierank> atomnuker: the binary is 55MB
[10:59:32 CET] <kierank> they don't need the space
[11:42:41 CET] <wm4> <kierank> It seems Chrome Embedded Framework lets you build ffmpeg and strip the licence flags <-for what purpose?
[11:42:55 CET] <wm4> because that makes you immediately think of supporting intentional license violations
[11:43:00 CET] <wm4> which would be Not Nice
[11:56:27 CET] <chouquette> Hello there, if some of you are at FOSDEM and would like to join the VideoLAN community dinner on Saturday evening, please get in touch with me :) (hugo at videolan.org)
[12:47:20 CET] <wm4> qsv sure has some horrible ifdeffery for a single vendor portable lib
[12:47:29 CET] <wm4> the libavcodec wrapper I mean
[14:48:48 CET] <jamrial_> michaelni: can you upload the sample attached in "fate: add id3v2 test" ?
[14:49:21 CET] <jamrial_> maybe rename it to id3v2_priv.mp3 since id3v2-test.mp3 is too generic, and the test is for the PRIV tag
[14:53:11 CET] <wm4> yeah with _priv it'd probably be better
[15:12:33 CET] <jamrial_> nevcairiel: do you mind if i push the ffprobe patch for coded_w/h? it's local, simple, and really innocuous, and will be gone in a year or two alongside avstream.codec
[15:13:05 CET] <michaelni> jamrial_, you can upload it yourself, ive just created an account and you should have permissions to upload samples
[15:51:34 CET] <SortaCore> er, latest ffmpeg doesn't have codec names before the hwaccels?
[16:10:38 CET] <wm4> so I guess I'll push my id3 and rtsp changes soon
[16:11:08 CET] <SortaCore> does rtsp have a useful timeout now?
[16:11:09 CET] <wm4> for id3 I suspect that there are some files around that expect pre-patch semantics, but didn't find any
[16:11:12 CET] <wm4> no
[16:11:15 CET] <SortaCore> rats
[16:11:27 CET] <wm4> well it does, just the option name is different
[16:11:34 CET] <wm4> which will be fixed in 2 years
[16:11:44 CET] <wm4> unless one of the bikeshedders go ahead and implement what they wanted
[16:12:01 CET] <SortaCore> now that's a term I'm not familiar with
[16:12:24 CET] <SortaCore> heh
[16:18:43 CET] <cone-268> ffmpeg 03Richard Shaffer 07master:4be6307cbf81: fate: add id3v2 test
[16:37:13 CET] <cone-268> ffmpeg 03Calvin Walton 07master:108958e43df0: librsvgdec: Fix frame clearing code
[16:37:31 CET] <atomnuker> kepstin: pushed, thanks
[16:37:46 CET] <kepstin> atomnuker: cheers!
[17:05:23 CET] <Chloe> atomnuker: how would you like me to credit you? Based on an idea from a previous patch by atomnuker or smth?
[17:05:48 CET] <atomnuker> yeah, just put a line in the description
[17:06:40 CET] <atomnuker> also you technically you finished the patch so "Based on an unfinished patch by atomnuker" is more accurate
[17:08:32 CET] <wm4> when will git have support for multiple author fields
[17:11:56 CET] <Chloe> atomnuker: thanks, Ill add it in :)
[17:16:24 CET] <atomnuker> wm4: when starts using a better sha version and does a 3way <everything> by default
[17:25:42 CET] <Chloe> adding in author renames would be nice too
[17:27:46 CET] <atomnuker> author renames? like authors changing names?
[17:28:08 CET] <atomnuker> or emails? I'd like that
[17:31:16 CET] <Chloe> atomnuker: both, but mostly authors changing names
[17:38:20 CET] <wm4> he's probably referencing the x264 incident
[17:38:43 CET] <atomnuker> well you don't have to change your name if you're wise like ^^^ is
[17:39:41 CET] <Chloe> atomnuker: it's more difficult if you dont have forethought like wm4 
[17:43:04 CET] <Chloe> atomnuker: also people change their names often, it seems kind of silly to not have a way to do it
[17:44:13 CET] <Chloe> wm4: I have no idea what that is
[17:48:16 CET] <atomnuker> Chloe: project leader of x264 changed his name/email and rewrote the git history
[17:49:01 CET] <wm4> he didn't just change name, he changed his gender
[17:49:27 CET] <wm4> and I think it wasn't officially explained or something, they just force pushed it, rewriting years or history
[18:43:47 CET] <Chloe> michaelni: yes I did build out of tree
[18:52:27 CET] <Chloe> michaelni: did you try a distclean first?
[19:24:55 CET] <michaelni> Chloe, let me retry
[19:27:12 CET] <SortaCore> what happened to all the hwaccels on master?
[19:28:37 CET] <wm4> SortaCore: nothing?
[19:29:27 CET] <SortaCore> h264_qsv, h264_cuvid and h264_nvenc hwaccels are gone
[19:29:33 CET] <SortaCore> although I see a h264_nvdec
[19:29:56 CET] <wm4> I doubt nvenc ever had a hwaccel, because it's an encoder
[19:30:14 CET] <wm4> the cuvid/qsv hwaccels are gone because they were just dummy which are not required anymore
[19:30:26 CET] <wm4> the actual hardware accelerated decoders are still there
[19:30:46 CET] <SortaCore> so if you're using one decoder and one encoder with nvenc, what's the process now?
[19:31:02 CET] <SortaCore> nvdec hwaccel for src, then create derived dxva2 for destination?
[19:31:57 CET] <wm4> exactly the same as before
[19:32:26 CET] <wm4> if you can use nvdec that's generally better than cuvid (IMO), although it lacks deint/scaling
[19:33:45 CET] <michaelni> Chloe, "mkdir new-dir && cd new-dir && ../configure && make -j12" reproduces the issue with the first patch on top of caaa40d2c67b1f4ebde368c859647af6f42f394a
[19:34:08 CET] <michaelni> "make distclean" isnt meaningfull as the build directory is empty
[19:40:18 CET] <atomnuker> kierank: this is horrible, its friday night, and brussels is pretty much dead, no kebab shops nearby, no italian restaurants, only very fancy places open with 0 people in them
[19:40:21 CET] <michaelni> Chloe, running make distclean in the parent directory seems to fix it but that really shouldnt be
[19:40:50 CET] <Chloe> I think I forgot to add the directories to the _list.c includes
[19:41:00 CET] <kierank> Do you want to come to "gist" with us?
[19:41:08 CET] <atomnuker> sandwiches in supermarkets are what's left and even then there's no more than a few, and even then I've never noticed how everything in brussels is turboexpensive
[19:41:40 CET] <wm4> heh how can it possibly be so dead if it's not even 8 PM?
[19:41:43 CET] <atomnuker> 5 euro for 100 grams of chocolate which would go for no more than 1.5 quids
[19:41:45 CET] <kierank> atomnuker: we are going to "gist"
[19:43:01 CET] <kierank> I have no idea what it is
[19:43:22 CET] <atomnuker> too late, had an unusual tuna sandwich for which I managed to aquire a taste for and my hotel is near ULB (for now, I'm going to the novotel after fosdem)
[20:00:23 CET] <Chloe> jamrial_: Should I add a rename for bsf to the next iteration of the set?
[20:00:30 CET] <Chloe> (with deprecation ofc)
[20:01:22 CET] <jamrial_> I have no real opinion about it. But I'd rather stop seeing two separate and conflicting sets doing the same thing in different ways, to be honest
[20:01:31 CET] <jamrial_> it's duplicate work
[20:01:59 CET] <Chloe> I dont know what prompted the other set tbh
[20:03:23 CET] <Chloe> the other set also only fixes half the issue, and doesnt allow for changing how codecs are stored in the future without another API change. With my changes you could theoretically reintroduce a registration API and keep the same iteration APIs without having to play with linked lists (since they're opaque)
[20:11:08 CET] <Chloe> what. why am I being told *why* I made a particular change
[20:22:30 CET] <nevcairiel> jamrial_: its fine to push that, i guess, i still think its a bit silly, but not going to block it
[20:24:21 CET] <jamrial_> ok
[20:40:37 CET] <peloverde> Anyone in contact with the homebrew ffmpeg maintainers? A few of their defaults look interesting. 
[20:41:35 CET] <Chloe> peloverde: you can fairly easily create an issue on their github page if you think they should be change
[20:41:47 CET] <cone-386> ffmpeg 03Zhong Li 07master:19b1d905b88d: ffprobe: Initialize coded_width/height
[20:44:42 CET] <Chloe> michaelni: I've updated the set, out of tree builds should just work now
[20:44:57 CET] <Chloe> also bsf api name patch added
[21:51:52 CET] <cone-386> ffmpeg 03James Almer 07master:94eb5505ad7b: ffprobe: remove usage of deprecation warning removal pragmas
[22:26:18 CET] <atomnuker> wm4: going to push the id3 and rstp stuff?
[22:31:17 CET] <wm4> needs some adjustment, so tomorrow
[00:00:00 CET] --- Sat Feb  3 2018