[Ffmpeg-devel-irc] ffmpeg-devel.log.20171227

[burek]

[00:06:53 CET] <BBB> whats funny is that in classic programming, youre taught to re-use existing code (like ff_update_block_index()) so that you have less bugs and more features, e.g. to suppose 420, I only have to implement it once; however, it appears that in this case, re-using existing code prevents us from doing exactly that. Why is that?
[00:07:48 CET] <BBB> s/suppose /support non-/
[00:12:16 CET] <Compn> i'm guessing because you cannot future proof the code 
[00:12:39 CET] <Compn> not 'you' as in you but maybe not even possible to do so... at least with speeds in mind
[00:13:09 CET] <durandal_1707> microoptimizations
[00:13:16 CET] <Compn> e.g. you write 10 bit code, and 12 bit comes out
[00:13:28 CET] <Compn> you write 16 bit and convert 8 bit to 16 bit and everyone complains :D
[00:13:53 CET] <Compn> so the answer is, we try to please everyone , support everything, and be the fastest
[00:14:14 CET] <Compn> the strategy has worked so far, ff is on every computer in the world now
[00:14:52 CET] <Compn> you would think microsoft would try to engineer a replacement
[00:41:03 CET] <Djfe> hardly worth it and licensing is quite the hassle, I guess ^^
[00:47:08 CET] <atomnuker> durandal_1707: working on atrac9
[00:47:27 CET] <atomnuker> laptop's running arch for now, though I'll likely reinstall in a month again
[00:47:45 CET] <atomnuker> because new kde version might finally be really usable... or maybe wlroots will be good enough
[00:48:13 CET] <atomnuker> also why the hell are encoder samplerates NOT sorted?
[00:49:00 CET] <atomnuker> I mean in an ascending order
[01:13:13 CET] <kierank> is anyone good with c preprocessor?
[01:15:04 CET] <kierank> #define FUNC3(a, b, c)  a ## _ ## b ## c
[01:15:04 CET] <kierank> #define FUNC2(a, b, c)  FUNC3(a, b, c)
[01:15:07 CET] <kierank> so what's the point of FUNC2?
[01:15:24 CET] <kierank> I want a_b_c
[01:16:27 CET] <jdarnley> make a macro a ## _ ## b ## _ ## c or put the underscore in b or c
[01:16:46 CET] <jdarnley> as for the duplicate, maybe to work around some old bug?
[01:17:03 CET] <sfan5> i think those setups are because you need to "expand" the arguments first
[01:17:09 CET] <sfan5> otherwise you end up with a_b_c (literally)
[01:49:51 CET] <kierank> jdarnley: a ## _ ## b ## _ ## c doesn't work
[01:50:12 CET] <kierank> ff_simple_idct_put_BIT_DEPTH_IN_IDCT_DEPTH
[01:50:20 CET] <kierank> doesn't evaluaethe the values
[01:52:33 CET] <jdarnley> Is that why there are two macros?  Because stuff like that doesn't work?
[01:52:56 CET] <jdarnley> like sfan5 suggested?
[01:53:08 CET] <nevcairiel> it is, if you want to contact strings you need to expand first, thats why its doubled
[01:53:15 CET] <kierank> I've done 
[01:53:16 CET] <kierank> #define FUNC4(a, b, c)  a ## _ ## b ## _ ## c
[01:53:16 CET] <kierank> #define FUNC5(a)  FUNC4(a, BIT_DEPTH, IN_IDCT_DEPTH)
[01:53:28 CET] <kierank> I have to make a new func for each one?
[01:53:36 CET] <jdarnley> Oh, other cpp macros?
[01:53:47 CET] <kierank> yes
[02:00:47 CET] <nevcairiel> yeah that wont work without another level, but you can use a generic macro to just expand those macros, i'm sure we have those somewhere
[02:11:41 CET] <BBB> kierank: yes, you need pre-expansion for that
[02:11:56 CET] <BBB> kierank: its kind of silly but I ran into that problem all the time also when working with these functions
[02:12:02 CET] <BBB> thats what the double macros are for :(
[02:12:13 CET] <kierank> BBB: jdarnley: sent what I have to the list
[02:12:25 CET] <BBB> woohoo
[02:12:39 CET] <kierank> just the templating stuff for now, not sure if it can be improved
[02:13:08 CET] <kierank> I made all the simple idct functions go from func_{bitdepth} to func_{in_bit_depth}_{out_bit_depth}
[03:45:53 CET] <wm4> tmm1: hm gcc doesn't warn
[03:46:35 CET] <wm4> but also doesn't emit a deprecation warning on use
[03:49:19 CET] <cone-741> ffmpeg 03wm4 07master:8f9024f2ca47: lavc: remove uneffective attribute_deprecated on enum
[03:49:28 CET] <wm4> wait, it's "ineffective"
[03:49:29 CET] <wm4> damn
[03:51:48 CET] <jamrial_> wm4: gcc 6 and newer let you deprecate specific enum values, but otherwise i don't think it's supported
[03:53:10 CET] <wm4> yaeh, I just marked them deprecated in the comment instead - clear enough
[04:09:18 CET] <tmm1> thx
[06:02:14 CET] <Compn> [jpeg2000 @ 0000000000608960] Could not find Jpeg2000 codestream atom.
[06:02:14 CET] <Compn> [image2 @ 0000000000462400] decoding for stream 0 failed
[06:02:15 CET] <Compn> cmon
[06:02:21 CET] <Compn> why u do dis
[06:04:03 CET] <Compn> any new devs want to fix jp2 decoding ? :)
[09:47:19 CET] <j-b> Good morning!
[10:54:10 CET] <durandal_1707> Good morning!
[11:12:01 CET] <cone-691> ffmpeg 03Paul B Mahol 07master:caacbfa773fc: avfilter/vf_convolve: cosmetics
[11:12:02 CET] <cone-691> ffmpeg 03Paul B Mahol 07master:87f148d52634: avfilter/vf_convolve: remove padding, its unused and not needed
[11:57:23 CET] <jdarnley> What?  Why does WSL need something special?  Can't you just run the standard Ubuntu compilers and the exe they will produce?
[11:57:51 CET] <nevcairiel> you can, but people want to use windows compilers
[11:58:33 CET] <jdarnley> Is that even possible?  I thought WSL was a ghetto that didn't allow access
[11:58:41 CET] <atomnuker> but windows compilers are junk, why would people want to use them?
[11:58:58 CET] <cone-691> ffmpeg 03Paul B Mahol 07master:8c9a91ac82b9: avfilter: add deconvolve filter
[11:59:01 CET] <nevcairiel> please keep your trolling to yourself
[11:59:10 CET] <JEEB> ugh
[11:59:27 CET] <JEEB> using windows mingw-w64 compilers from a "proper" *nix
[11:59:29 CET] <JEEB> welp
[11:59:42 CET] <JEEB> is cross-prefix understanding so hard
[11:59:46 CET] <jdarnley> Oh, that person even wants to compile for ARM
[12:00:46 CET] <nevcairiel> changing configure to directly reference the msvc files as .exe is probably fine though, thats what they are named anyway and would allow wsl to find them
[12:01:04 CET] <jdarnley> Oh god.  The guy doesn't even know what he wants.  "get yasm and gas-preprocessor"
[12:01:12 CET] <JEEB> LOL
[12:02:16 CET] <atomnuker> nevcairiel: fine, they're junk for c because they're c++, that's a fact
[12:02:21 CET] <atomnuker> so why do people use them?
[12:04:00 CET] <jdarnley> If anyone cares this might be the referenced patch https://github.com/Microsoft/FFmpegInterop/blob/gillesk/wsl/0001-Updating-scripts-to-run-under-WSL.patch
[12:04:13 CET] <jdarnley> And with that I am done looking down that rabbit hole
[12:04:18 CET] <nevcairiel> I use them for debugging, because its the only way to directly do step-by-step debugging through my calling code
[12:04:38 CET] <nevcairiel> yeah most those changes just directly append the .exe so wsl can find it
[12:05:12 CET] <nevcairiel> not sure what the CC_IDENT escaping is about
[12:05:39 CET] <jdarnley> I thought it was stripping carriage returns
[12:05:50 CET] <jdarnley> '\r'
[12:21:19 CET] <robertfoss_> heya
[12:22:09 CET] <robertfoss_> I'm building ffmpeg and having some issues enabling the right flags for enabling png codec + muxer
[12:22:30 CET] <robertfoss_> I am running ./configure with --enable-zlib
[12:24:42 CET] <durandal_1707> atomnuker: are you gonna write decoder, have you even started?
[12:26:09 CET] <robertfoss_> https://hastebin.com/gigorareme.pas
[12:26:56 CET] <atomnuker> durandal_1707: I'VE STARTED
[12:33:40 CET] <cone-691> ffmpeg 03Paul B Mahol 07master:aff167847754: configure: note (de)convolve filter dependency
[12:39:44 CET] <durandal_1707> atomnuker: STOP USING CAPS!!
[12:39:53 CET] <jdarnley> robertfoss_: 1 - this should be in #ffmpeg. 2 - i think ffmpeg can't determine the properties of the input correctly.
[12:47:06 CET] <robertfoss_> jdarnley: ack, moving over :)
[13:54:11 CET] <durandal_1707> Compn: dont be lazy, open bug report
[14:53:50 CET] <robertfoss_> jdarnley: not getting any replies, whom should I poke?
[15:21:27 CET] <Compn> durandal_1707 : oh yeah, good idea
[15:31:25 CET] <durandal_1707> michaelni: if you do not wish to relicense geq filter i will write new generic expresssion filter
[15:33:09 CET] <durandal_1707> i will also write filter which calculates frame entropy
[15:34:19 CET] <atomnuker> yes, such a filter would be awesome
[15:34:52 CET] <michaelni> durandal_1707, why do you care about the license ?
[15:36:07 CET] <durandal_1707> michaelni: trivial filters shoukd be lgpl
[15:52:50 CET] <thardin> I'm surprised ffmpeg isnt agpl yet
[15:53:24 CET] <thardin> given how often changes to it are hidden in ~the cloud~
[15:54:12 CET] <durandal_1707> what have i missed?
[15:54:26 CET] <thardin> youtube?
[15:54:55 CET] <durandal_1707> irc logs
[15:55:04 CET] <thardin> oh
[15:55:33 CET] <thardin> i ignorr joins+parts
[15:56:17 CET] <durandal_1707> thardin: from my last message?
[15:57:37 CET] <thardin> only me suggesting ffmpeg's clis should be agpl since changes to them are hidden in the cloud
[15:58:38 CET] <thardin> and indirectly changes to the libs. but they may well stay lgpl
[15:58:45 CET] <thardin> imo
[15:59:58 CET] <thardin> may be more relevant for ffmbc
[16:01:00 CET] <durandal_1707> saste: where is my paycheck?
[17:50:17 CET] <atomnuker> takes time for spi to do it, I think they arrived in february last year
[17:50:24 CET] <atomnuker> *this year
[18:25:05 CET] <kierank> jdarnley: any thoughts on my patch
[18:28:01 CET] <jdarnley> I haven't looked at it indepth but I was impressed that you managed to do it with so few changes.
[18:49:06 CET] <kierank> jdarnley: i only did the C
[18:49:14 CET] <kierank> I guess only you and BBB would understand that code
[18:49:28 CET] <kierank> but I do wonder if fundamentally there is a better way of doing this
[19:39:18 CET] <cone-691> ffmpeg 03James Almer 07master:efb63e4316c2: configure: add missing avcodec dep to avfilter for de/convolve filters
[19:48:15 CET] <kierank> michaelni: apart from it breaking that easily fixable testprog, do you have any objections?
[21:04:49 CET] <michaelni> durandal_1707, if you need libavfilter/vf_geq.c under LGPL iam not against it being relicensed. But more generally for things that are more complex i think we should not relicense GPL->LGPL for free. That is if theres a company that needs LGPL, they can as well pay the authors and donate to the project or some good cause or something
[21:07:50 CET] <Chloe> where does project money go anyway
[21:09:08 CET] <Compn> Chloe : into ffmpeg fund , gets paid out to refund developers costs of going to conventions. tshirt printing and shipping too...
[21:09:22 CET] <Compn> we havent used it for much else except that i dont think
[21:09:38 CET] <Compn> maybe bought hardware for one developer once or something. i know we had to buy some server hardware once
[21:09:51 CET] <Compn> but that was with mplayer project too :)
[21:10:35 CET] <Compn> i like michaels idea of getting paid to gpl>lgpl things
[21:11:01 CET] <Compn> everyone needs money to survive :\
[21:11:19 CET] <Compn> these companies make billions of dollars on ffmpeg's back
[21:11:45 CET] <Compn> some companies give back, others do not
[21:13:07 CET] <Chloe> can I submit patches as gpl and ffmpeg pays me to lgpl them
[21:14:52 CET] <Compn> you can submit new gpl licensed features and then wait for companies to pay you to lgpl them
[21:15:09 CET] <Compn> but i dont think we will accept gpl/lgpl mixed code , too complicated
[21:15:20 CET] <Chloe> shame
[21:15:21 CET] <Compn> it works better for entire file licenses
[21:15:34 CET] <Chloe> ffmpeg should do more bounties
[21:15:36 CET] <Compn> also ffmpeg would not really pay anyone to relicense i dont think
[21:18:42 CET] <Compn> Chloe : but there are always opportunities. adding yourself to consultants page may land you some ffmpeg related jobs
[21:18:51 CET] <kierank> Chloe: vlc has many ffmpeg bounties already
[21:19:07 CET] <Compn> http://ffmpeg.org/consulting.html
[21:19:08 CET] <kierank> Chloe: https://wiki.videolan.org/Bounties/
[21:19:24 CET] <Compn> ffmpeg has some bounties in the trac as well ? maybe ?
[21:20:03 CET] <Chloe> there was a single bounty in the trac i think
[21:20:26 CET] <Chloe> I guess I need to suck it up and learn to reverse codecs if I want to earn any money
[21:20:37 CET] <Chloe> durandal_1707: what do you use for reversing? IDA?
[21:20:52 CET] <kierank> voxware metasound is doable i think
[21:21:06 CET] <durandal_1707> there is already decoder
[21:21:11 CET] <durandal_1707> buggy
[21:21:16 CET] <durandal_1707> incomplete
[21:22:05 CET] <durandal_1707> ClearVideo is incomplete too
[21:22:05 CET] Action: Compn better update his ffmpeg before reporting bug
[21:22:19 CET] <iive> why is vlc offerring bounty for aac encoder that is at least as good as nero libfaac ?
[21:23:00 CET] <Chloe> iive: 'Past bounties'
[21:23:00 CET] <Compn> nero libfaac has bad license ?
[21:23:06 CET] <durandal_1707> Chloe: nobody normal can afford ida
[21:23:35 CET] <iive> gah
[21:23:40 CET] <Chloe> durandal_1707: well I know you've reversed a bunch of stuff so I wondered if you had got yourself a copy
[21:23:48 CET] <Chloe> werent kodi handing out copies
[21:24:15 CET] <Compn> kodi hands out a lot of useful stuff to developers
[21:24:25 CET] <durandal_1707> nope,  who gave you such nonsense stuff
[21:24:26 CET] <Compn> i recommend asking them if you need a tool or hardware
[21:24:34 CET] <Compn> durandal_1707 : ask kodi for what you need.
[21:24:38 CET] <Compn> they will give you 
[21:25:12 CET] <durandal_1707> Chloe: there is free dissasembler
[21:25:30 CET] <durandal_1707> with some lick you can dechiper assembly
[21:25:42 CET] <durandal_1707> *luck
[21:26:55 CET] <durandal_1707> and free decompiler on www
[21:27:27 CET] <durandal_1707> but ida decompiler is higher quality
[21:30:16 CET] <kierank> hopper is also good
[21:30:21 CET] <kierank> on linux and mac
[21:31:07 CET] <durandal_1707> for legit ida you need to pay little fortune
[21:32:01 CET] <Chloe> i like hopper for reversing objc but need something on windows probably
[21:36:38 CET] <Compn> there you go, durandal_1707 ^
[21:36:39 CET] <durandal_1707> Compn: will you upload sample?
[21:36:44 CET] <Compn> i posted sample url
[21:36:51 CET] <Compn> https://chroniclingamerica.loc.gov/lccn/sn85032933/1908-11-14/ed-1/seq-1.jp2
[21:36:55 CET] <Compn> right there and in trac :P
[21:37:08 CET] <Compn> it was 3mb , 500k too big to be uploaded in trac :(
[21:38:09 CET] <Compn> see now this ticket, if valid, would be good for new developer to figure out whats up with j2k 
[21:38:18 CET] <Compn> and then add the required code to fix it
[21:38:49 CET] <Compn> durandal_1707 or carl could probably fix this in a day or two. but the point is i'm trying to find easy bugs not for them, but for new contributors
[21:39:11 CET] <llogan> someone on twitter wants to know if we support "all dolby formats"? i know nothing of that so what should I tell them?
[21:39:16 CET] <durandal_1707> i work on j2k only if paid
[21:39:24 CET] <llogan> https://twitter.com/WinIsBest/status/946009605544513536
[21:39:47 CET] <Compn> llogan : theres probably some dolby hardware spdif stuff we dont support
[21:39:56 CET] <Compn> llogan : dolby has a lot of tech that no one really plays with
[21:40:13 CET] <Compn> i dont think we support SACD stuff either
[21:40:33 CET] <durandal_1707> Compn: whats that?
[21:41:15 CET] <Compn> super audio cd > https://en.wikipedia.org/wiki/Super_Audio_CD
[21:42:53 CET] <Compn> oh i dont see dolby involved with SACD so nevermind about that format
[21:43:07 CET] <durandal_1707> Compn: dsd is supported and thats in sacd
[21:43:55 CET] <Compn> hah dolby uses ffmpeg too
[21:44:12 CET] <Compn> This software uses libraries from the FFmpeg project under the LGPLv2.1
[21:44:17 CET] <durandal_1707> llogan: dts:x, specifically object channels are not supported at all
[21:44:21 CET] <Compn> Dolby CineAsset mastering software suite 
[21:44:24 CET] <rcombs> no Dolby Vision support either
[21:44:30 CET] <rcombs> or any Atmos
[21:44:40 CET] <durandal_1707> yes atmos
[21:44:41 CET] <rcombs> or eac3 with dependent substreams
[21:44:45 CET] <durandal_1707> and dts x
[21:45:01 CET] <Compn> what are the dolby audio streams in dcp ? do we support those ?
[21:45:02 CET] <durandal_1707> but dts is not dolby
[21:45:20 CET] <rcombs> (fun fact: an atmos software decoder exists, and I will say little else of it here)
[21:45:29 CET] <Compn> thats atmos i see i see
[21:46:13 CET] <Compn> llogan : you can always use twitter to ask for dolby samples we dont yet support :)
[21:46:21 CET] <durandal_1707> rcombs: where? say more
[21:46:46 CET] <durandal_1707> rcombs: is it open source?
[21:47:39 CET] <kierank> there is an atmos spec
[21:47:47 CET] <kierank> and I heard there is an atmos software decoder, yes
[21:49:55 CET] <durandal_1707> we will get sued if we do it, perfect oportunity to kill project
[21:50:36 CET] <llogan> how about "Dolby Atmos and Dolby Vision are not currently supported."?
[21:51:02 CET] <durandal_1707> lgtm
[21:51:08 CET] <llogan> thanks
[21:52:57 CET] <llogan> another twitter guy wants to know if any ffmpeg devs are at 34CC (Chaos Communication Congress)
[21:53:20 CET] <llogan> 34c4 i guess is the actual tag
[21:53:36 CET] <durandal_1707> no, we do not tolerate chaos
[21:54:13 CET] <durandal_1707> but there are ffmpeg pro users there for sure
[21:58:02 CET] <Compn> llogan : thats in berlin
[21:58:05 CET] <Compn> iirc
[21:58:10 CET] <Compn> thilo might be ?
[21:58:22 CET] <Compn> i dont remember him saying he was going  thou
[21:58:43 CET] <llogan> i told him i didn't know but look for someone wearing ffmpeg shirt
[21:58:46 CET] <Compn> thilo is the only ff dev i know in berlin so :P
[21:58:47 CET] <Compn> ehehe
[22:01:12 CET] <kierank> I was thinking of going
[22:01:22 CET] <Chloe> I went last year, was pretty good
[22:01:45 CET] <durandal_1707> what is it about?
[22:01:59 CET] <kierank> hacking conf
[22:02:03 CET] <kierank> it's too political for me
[22:02:11 CET] <llogan> too far for me. like everything.
[22:02:47 CET] <durandal_1707> kierank: what kind of political?
[22:02:58 CET] <kierank> the kind of michaelni signature politics
[22:03:01 CET] <Compn> durandal_1707 : hangout meetup for hackers and open source and computer politics. like defcon except less corporate / nsa
[22:03:11 CET] <durandal_1707> lol
[22:04:08 CET] <Compn> The Chaos Communication Congress is the Chaos Computer Club's (CCC) annual symposium and hacker party. During four days between Christmas and New Years Eve, thousands of hackers, technology freaks, artists, and utopians get together in Leipzig to communicate, learn from each other, and party
[22:04:26 CET] <Chloe> I just avoided all the political stuff and only went to technical talks
[22:04:49 CET] <sfan5> Compn: it's in Leipzig this year by the way
[22:04:53 CET] <Compn> i see :P
[22:04:57 CET] <Compn> i dont know where that is
[22:05:43 CET] <Compn> like an hour or two south of berlin
[22:05:45 CET] <sfan5> not very far from Berlin
[22:15:19 CET] <Compn> durandal_1707 : berlin is very friendly to young computer hackers, seems like a big crowd there now
[22:15:34 CET] <Compn> go if you get the chance
[22:15:48 CET] <durandal_1707> im old fart
[22:30:12 CET] <Chloe> should do an age survey on all the ffmpeg people
[22:31:26 CET] <durandal_1707> average age: 67
[23:31:40 CET] <atomnuker> llogan: if I knew it started today a month ago I'd have gotten tickets
[23:32:02 CET] <atomnuker> maybe 2, since flight prices are through the roof
[23:32:40 CET] <atomnuker> I actually looked it up a few hours ago just to check if it really was today and had a laugh
[23:32:51 CET] <atomnuker> their wiki has a page with dating ads
[23:32:58 CET] <atomnuker> as well as a dating manual
[23:33:24 CET] <atomnuker> https://events.ccc.de/congress/2017/wiki/index.php/Session:Daygame_-_meeting_girls_(and_boys)_in_daily_situations
[23:33:29 CET] <atomnuker> https://events.ccc.de/congress/2017/wiki/index.php/Dating
[23:35:36 CET] <llogan> atomnuker: yeah, I saw that too when i was trying to figure out what ccc was. first link in search. entertaining. i know how it feels to pay for high ticket prices. went to ireland in May. friends in "lower 48" paid 50% less.
[23:44:29 CET] <wm4> " Looking for someone who can help me to get Rust running on my STM32 hardware - Contact me @ Twitter or DeepCyber assembly."
[23:44:32 CET] <wm4> sexy
[23:44:50 CET] <wm4> " Looking for someone having/doing NFC Tag Implants"
[23:45:47 CET] <wm4> " looking for 420, contact over jabber: nand at jabb3r.org or the mail in my profile [...]"
[23:46:00 CET] <wm4> oh failed to censor the first email address
[23:46:54 CET] <kierank> ?
[23:47:28 CET] <wm4> from the ccc "Dating" page
[23:47:35 CET] <wm4> which atomnuker linked above
[23:48:32 CET] <kierank> ah
[23:48:53 CET] <wm4> " a couple, energetic, kinda very funny and looking for a girl and/or a boy to storm through the congress nights. And you may wanna be a bit kinky. "
[23:48:57 CET] <wm4> that's on the same page
[23:48:58 CET] <wm4> ...
[23:49:10 CET] <kierank> nothing wrong with that
[23:49:12 CET] <kierank> a free country
[23:49:33 CET] <wm4> "Two handsome otters (20 & 22, "twink") seek an additional otter for some new experience. "
[23:49:42 CET] <wm4> I don't even know what that's code for
[23:50:10 CET] <Chloe> 'otter' is like 'bear', terms from gay culture afaik
[23:50:17 CET] <Chloe> I think
[23:51:33 CET] <nevcairiel> <atomnuker> llogan: if I knew it started today a month ago I'd have gotten tickets <-- its the exact same date every year, so consider yourself in the known for the future? :)
[23:57:34 CET] <Compn> lol carl fixed jp2k issue
[00:00:00 CET] --- Thu Dec 28 2017