[Ffmpeg-devel-irc] ffmpeg-devel.log.20170222

burek burek021 at gmail.com
Thu Feb 23 03:05:02 EET 2017


[00:10:44 CET] <cone-751> ffmpeg 03Carl Eugen Hoyos 07master:2ac381088d05: lavd/opengl_enc: Support BGR48.
[00:13:37 CET] <cone-751> ffmpeg 03Alex Converse 07master:dde1bf074ca6: aacdec: When ignoring a PCE restore the previous config
[00:22:25 CET] <cone-751> ffmpeg 03James Almer 07master:d04e0a6bcffe: avcodec/qdrw: don't overwrite bpp when checking its value
[01:16:22 CET] <cone-751> ffmpeg 03James Almer 07master:1ad60e4e70ec: avformat/matroskaenc: don't write DisplayUnit with value Unknown on WebM files
[01:18:46 CET] <cone-751> ffmpeg 03James Almer 07release/3.2:36fff6c75459: avformat/matroskaenc: don't write DisplayUnit with value Unknown on WebM files
[02:45:33 CET] <cone-751> ffmpeg 03Michael Niedermayer 07master:24f6559bd241: avcodec/rv34: Fix runtime error: signed integer overflow: -2 + -2147483648 cannot be represented in type 'int'
[02:45:34 CET] <cone-751> ffmpeg 03Michael Niedermayer 07master:92188c8f57bf: avcodec/dcadsp: Fix runtime error: signed integer overflow: 394625024 * 8 cannot be represented in type 'int'
[02:45:35 CET] <cone-751> ffmpeg 03Michael Niedermayer 07master:6179dc8aa7e5: avcodec/mpeg4video: Fix runtime error: left shift of negative value
[02:45:36 CET] <cone-751> ffmpeg 03Michael Niedermayer 07master:c9e3952b82e5: avcodec/rv34: Forward error from rv34_decode_mv()
[02:47:07 CET] <atomnuker> RiCON: ok, maybe your gcc isn't crazy
[02:47:29 CET] <atomnuker> my fuzzing program doesn't consider K == 0 which can happen and does fail here
[02:51:41 CET] <RiCON> i'd figured out my gcc wasn't crazy when nevcairiel's failed too :/
[03:09:33 CET] <iive> RiCON: have you tried your sample with linux build?
[03:13:45 CET] <iive> have you provided sample to atomnuker ?
[03:14:02 CET] <RiCON> not yet, only have an atom with it
[03:14:06 CET] <RiCON> and yes, first thing i did
[03:15:05 CET] <RiCON> https://i.fsbn.eu/instantcrash.flac
[03:15:39 CET] <RiCON> i assumed my own linux build wasn't needed since atomnuker couldn't reproduce it
[03:15:51 CET] <iive> what  was your command line?
[03:16:10 CET] <iive> no, it is not needed.
[03:16:19 CET] <atomnuker> I still can't reproduce it, though I did manage to hit a failure with another sample at 2.5 ms
[03:16:20 CET] <RiCON> plain "ffmpeg -i http://i.fsbn.eu/instantcrash.flac out.opus"
[03:16:29 CET] <iive> i just want to know if it has been tested.
[03:16:31 CET] <RiCON> with -strict -2
[03:17:23 CET] <iive> will test tomorrow
[03:17:27 CET] <iive> see ya
[03:20:49 CET] <iive> i actually do have recent git build and no crash or assert with it.
[06:27:08 CET] <FishPencil> Why does '#include "libavutil/thread.h"' from a libavdevice header results in: src/compat/w32pthreads.h:80:20: error: invalid conversion from void* to pthread_t* [-fpermissive]
[06:40:22 CET] <wm4> FishPencil: where does this happen?
[06:41:38 CET] <FishPencil> wm4: Working on a patch, so it isn't happening in prod
[06:42:46 CET] <wm4> sounds like it's in C++ code
[06:43:22 CET] <FishPencil> wm4: You're correct, it's decklink which is C++.
[06:44:05 CET] <wm4> it does #include <pthread.h>
[06:47:50 CET] <FishPencil> wm4: I'm trying to tie into thread.h so that the pthread emulation is present for Windows when no pthread lib is included
[06:48:39 CET] <wm4> w32pthreads.h was obviously written with C only in mind
[06:48:44 CET] <wm4> so you need to adjust it
[06:48:45 CET] <FishPencil> Even extern "C" {#include "libavutil/thread.h"} doesn't work
[06:48:58 CET] <wm4> extern "C" only changes name mangling, not the language
[06:50:16 CET] <FishPencil> Is simply doing -fpermissive a bad idea?
[06:53:00 CET] <wm4> yes
[06:55:41 CET] <FishPencil> So how should this be tackled? Decklink doesn't actually depend on pthread, but the avdevice within FFmpeg does. To remove the dependency, should another emulation for pthread be created to work with Decklink?
[06:56:19 CET] <wm4> wouldn't it be a better idea to adjust the existing emulation
[06:57:12 CET] <FishPencil> You mean fix w32pthreads.h?
[06:58:43 CET] <wm4> yes
[07:46:28 CET] <nevcairiel> c++ needs a lot more explicit casts that C doesnt need
[07:46:35 CET] <nevcairiel> so those would need to be added
[11:11:45 CET] <durandal_1707> kierank: do you have dnxhr specification?
[11:12:17 CET] <kierank> dunno what is spec number
[11:12:22 CET] <kierank> i have plenty of hate apparently
[11:12:53 CET] <durandal_1707> hmm?
[11:33:55 CET] <kierank> ffmpeg.org down for me
[11:35:33 CET] <mateo`> same here
[12:10:50 CET] <durandal_1707> shit everything is down
[12:11:29 CET] <BtbN> trac works
[12:12:04 CET] <j-b> git works
[12:25:24 CET] <durandal_1707> samples are down they are essential
[12:25:26 CET] <kierank> this is what happens when you have conspiracy theorists calling the shots instead of getting vlc to host
[12:28:57 CET] <mateo`> what was the reason why we did not choose to depend on vlc for the hosting ?
[12:29:22 CET] <kierank> usual conspiracy theory "vlc may fork ffmpeg" crap
[12:30:21 CET] <durandal_1707> now im gonna do sonething else and ffmpeg community will suffer
[12:33:39 CET] <kierank> durandal_1707: you spread such hate
[12:34:35 CET] <mateo`> even if they fork, I don't see them removing credentials, taking over the website and burning everything
[12:34:55 CET] <durandal_1707> what? im completely inoccent
[12:35:58 CET] <durandal_1707> fixing dnxhr will need to wait
[12:36:34 CET] <durandal_1707> i couldnt get anything free to encode to cid 1271
[12:37:19 CET] <durandal_1707> and not going to buy expensive software to encode such samples
[12:51:13 CET] <ubitux> kierank: what are you talking about? videolan doesn't offer all the services we currently use, we already talked about that
[12:51:28 CET] <ubitux> cut the BS a few minutes, it's not because ffmpeg.org is down one day that we should shit on our infrastructure, we also suffer tech issues sometimes from videolan
[12:51:44 CET] <ubitux> you're more toxic than Carl, that's quite a performance
[12:52:25 CET] <kierank> I don't need to take lessons on toxicity from you 
[12:54:00 CET] <ubitux> yeah you don't, you're mastering it already
[12:57:00 CET] <kierank> Don't worry I have a lot to learn after your masterpiece:
[12:57:01 CET] <kierank> <kierank> ubitux: so it's fine to have racists, sexists, homophobes in FFmpeg just as long as they write patches?
[12:57:01 CET] <kierank> <ubitux> could be ok yes
[12:57:20 CET] <ubitux> and then you quit before i could end my sentence
[12:57:50 CET] <kierank> I wasn't the only one to quit because of that statement
[13:00:03 CET] <ubitux> we do not want to have any racists/sexists/homophobes statements or behaviour in FFmpeg but that's not our goal to do the police and check ppls background
[13:00:17 CET] <ubitux> just like we do not want to be involved in geopolitic
[13:00:18 CET] <kierank> stop digging
[13:01:02 CET] <ubitux> i don't want to care what someone is voting for, but i do care about ppl exhibiting such behaviour
[13:01:27 CET] <ubitux> that was the rationale of the statement before you exploded in an emotionnal and irrationnal furry
[13:01:27 CET] <kierank> FFTempleOS
[13:01:30 CET] <kierank> have fun with that
[13:01:40 CET] <kierank> this is what FFmpeg is becoming 
[13:02:51 CET] <ubitux> then fork FFmpeg and add in your policy that every developer must be a Sanders voter
[13:03:00 CET] <JEEB> tl;dr all infra has issues
[13:03:13 CET] <JEEB> you just need to calculate the pros/cons of everything and go on with that
[13:03:14 CET] <JEEB> :V
[13:03:32 CET] <durandal_1707> stop fight
[13:03:35 CET] <iive> there are infinite ways to divide people
[13:04:06 CET] <iive> not so many to unite them.
[13:04:16 CET] <kierank> it's already policy that you have to be a conspiracy theorist to be on ffmpeg
[13:05:17 CET] <kierank> I would link to all the conspiracy crap on ffmpeg-devel but I can't :)
[13:05:18 CET] <durandal_1707> ugh, im not spreading any theories. im just coding
[13:05:24 CET] <kierank> ok
[13:05:59 CET] <durandal_1707> i just dislike some carl patches
[13:07:10 CET] Action: iive facepalms
[13:07:19 CET] <durandal_1707> and why is oss fuzz elite groupation?
[13:07:29 CET] <kierank> because ffmpeg-security
[13:07:39 CET] <kierank> that is michael and his friends club
[13:08:03 CET] <durandal_1707> then let we become his friends
[13:09:34 CET] <ubitux> security is all about trust
[13:09:49 CET] <ubitux> kierank: so i'm assuming you're for full-disclosure?
[13:10:10 CET] <kierank> no, I'm for allowing more people on the security ML
[13:10:24 CET] <kierank> because lots of people have access to coverity, lots have git push
[13:10:40 CET] <ubitux> ask for it? i remember you said you didn't care anymore
[13:11:19 CET] <kierank> because whenever someone wants to do something in this project a wall of bureaucracy is put to protect the status quo
[13:11:32 CET] <kierank> bureaucracy that is made up on the fly
[13:11:41 CET] <ubitux> i asked in private on irc when i wanted it
[13:11:45 CET] <ubitux> really complex bureaucracy
[13:11:55 CET] <kierank> and look what happened when wm4 and I asked
[13:12:04 CET] <ubitux> i'm not in your private conversation
[13:12:07 CET] <kierank> (gain can't link to it.)
[13:12:13 CET] <kierank> was on public irc and on ml
[13:12:21 CET] <kierank> and then bureaucracy was thrown up
[13:14:08 CET] <ubitux> none of you are subscribed yet?
[13:17:25 CET] <iive> maybe they are, but there haven't been any mails yet.
[13:17:59 CET] <wm4> lol drama
[13:18:23 CET] <wm4> it turned out that ffmpeg-security isn't a mailing list, but an alias for michaelni's private mail, right?
[13:18:37 CET] <kierank> michael clement carl and reimar
[13:20:20 CET] <michaelni> "<kierank> that is michael and his friends club", this is slander, please do not do this again
[13:20:38 CET] <kierank> michaelni: why were they accepted without a vote yet wm4 and I need a vote
[13:20:53 CET] <wm4> michaelni: how is it slander
[13:20:58 CET] <kierank> 12:11 PM <"ubitux> i asked in private on irc when i wanted it
[13:20:59 CET] <kierank> 12:11 PM <"ubitux> really complex bureaucracy
[13:21:00 CET] <michaelni> kierank, because there are no rules sadly, 
[13:21:16 CET] <michaelni> i want there to be clear rules on who should be on the alias
[13:21:27 CET] <kierank> so when ubitux asked why didn't you post on the ml
[13:21:29 CET] <kierank> like you did for us
[13:21:50 CET] <ubitux> because i didn't make a fuss out of it i guess
[13:23:01 CET] <ubitux> it was years ago btw
[13:23:13 CET] <kierank> anyway I have proved all my points here so I consider this discussion over
[13:23:21 CET] <kierank> tune in next week for another episode
[13:23:24 CET] <ubitux> you were wrong on all of them so that's cool
[13:23:40 CET] <kierank> next weeks will be "why can certain people push without posting patches to the ML"
[13:23:51 CET] <kierank> same shows repeated every month on ffmpeg-devel!
[13:24:10 CET] <iive> and why is this an issue?
[13:25:39 CET] <wm4> I don't really get why people can push buggy shit without reviews, while my merge commits must solve EVERY ISSUE EVER
[13:26:12 CET] <wm4> I don't be misled by the trollishness of my statement, because it's actually true
[13:26:26 CET] <ubitux> well, it's a large change, so if the review raises regression they should be fixed?
[13:26:43 CET] <kierank> wm4: I look forward to hearing the arbitrary rules that will appear now
[13:27:22 CET] <michaelni> if a commit is buggy it should be reported to the author and fixed or reverted
[13:27:54 CET] <nevcairiel> maintainers can push to their maintained components, thats a rule that existed forever, if you dont like it suggest a policy change
[13:29:52 CET] <iive> i really liked ffmpeg much more, when there were less rules and people were trying to be nice
[13:30:21 CET] <wm4> "people were trying to be nice" when was that ever the case
[13:30:42 CET] <durandal_1707> in 2000?
[13:31:21 CET] <ubitux> iive: i doubt FFmpeg ever was a happy fest of nice people
[13:31:52 CET] <iive> it was for a few years after the fork
[13:32:07 CET] <ubitux> no
[13:32:38 CET] <durandal_1707> it was michaelni only
[13:32:55 CET] <ubitux> depressed people are not nice people iive :p
[13:33:47 CET] <iive> i said "trying to be nice"
[14:00:49 CET] <cone-439> ffmpeg 03Michael Niedermayer 07master:c93a07070a36: MAINTAINERS: Add ffmpeg-security alias members
[14:01:32 CET] <wm4> apparently michaelni is the maintainer of MAINTAINERS
[14:01:46 CET] <wm4> a consequence is that it basically enables him to do whatever he pleases
[14:02:07 CET] <wm4> (such as adding himself for components he'd like to change without going through a review process)
[14:02:17 CET] <michaelni> that was a patch on the mailinglist anyone could comment on
[14:02:58 CET] <wm4> hm ok
[14:06:59 CET] <iive> i actually miss the time when michael was ODFL and
[14:07:32 CET] <iive> ffmpeg was his pet project, where everybody who want to play could join him.
[14:07:47 CET] <wm4> was it ever "his"
[14:08:24 CET] <ismail> any idea why ffmpeg.org down?
[14:08:32 CET] <iive> it's quite horrible to see people arguing... "I want that toy", "He took my toy" "he doesn't want to give me that toy"
[14:11:06 CET] <wm4> iive: even if it was "his" (ignoring the fork situation etc.), he stepped down as "leader" (although it didn't have any real effect other than him stopping merges)
[14:11:12 CET] <AndrewMock> site down
[14:11:30 CET] <wm4> yes we're reminded every 5 minutes
[14:11:44 CET] <AndrewMock> k sry
[14:11:55 CET] <ismail> well add it to topic maybe then
[14:13:35 CET] <iive> wm4: i think that his stepping down was a mistake. It only balooned conflicts inside ffmpeg.
[14:13:39 CET] <JEEB> daily movenc sanity check removal
[14:13:41 CET] <JEEB> https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/movenc.c#L900
[14:14:01 CET] <JEEB> (negative CTS offsets makes that check incorrect)
[14:15:16 CET] <wm4> iive: I don't think they got more or less, nothing changed
[14:15:24 CET] <wm4> his stepping down is a farce
[14:16:46 CET] <wm4> JEEB: wow I hope that isn't asserting on input data?
[14:17:48 CET] <iive> wm4: well, he is still the man doing almost half of the work that have to be done.
[14:19:34 CET] <JEEB> wm4: pretty much on input to the muxer, yes
[14:19:44 CET] <iive> and yes, he stepped down because he saw hostility towards him increasing. It did not decrease the hostility, it just was not focused on him, thus the balooning.
[14:19:47 CET] <JEEB> I noticed during testing I was getting crashes when the fragments were getting written :D
[14:20:31 CET] <JEEB> (that happens with multiple b-frame enabled video tracks in a single movenc.c mux with negative cts offsets + fragments)
[14:20:51 CET] <JEEB> Shit happened with video (3): (next_dts - cluster_dts): -400000, cluster_dts: 49600000, next_dts: 49200000.
[14:20:54 CET] <JEEB> stuff like this
[14:39:42 CET] <sfan5> is it just me or is ffmpeg doing testing for openssl completely wrong? (non-pkgconfig path)
[14:39:53 CET] <sfan5> it tests for the symbol SSL_library_init
[14:40:07 CET] <sfan5> which isn't present in recent openssl versions because ./include/openssl/ssl.h:# define SSL_library_init() OPENSSL_init_ssl(0, NULL)
[14:40:30 CET] <wm4> iive: maybe the main issue is that we don't have a proper project leader who resolves these obnoxious shit-flinging mud fights
[14:42:29 CET] <jkqxz> sfan5:  ? <http://git.videolan.org/?p=ffmpeg.git;a=blobdiff;f=configure;h=481f69244bc50fdcfcd000a998a6ea31d51fc59f;hp=1e6834f2a99e78b78d3e29d9760ae1558666cca2;hb=798c6ecce50f26a53d48e2577a34fabe46b32eb9;hpb=0cfd6ccedeaa7fe2258c7590a754e3c1a4d7936f>
[14:42:51 CET] <sfan5> that only applies if you have pkgconfig
[14:43:23 CET] <sfan5> a further
[14:43:23 CET] <sfan5> +                               check_lib openssl/ssl.h OPENSSL_init_ssl -lssl -lcrypto ||
[14:43:23 CET] <sfan5> is needed to support openssl 1.1.0 without pkg-config
[14:45:35 CET] <durandal_1707> iive: its very easy to be leader and do most important task when you block others in doing same
[14:47:33 CET] <jkqxz> Hmm, why does it even have a non-pkgconfig path there at all?  That seems like a bizarre legacy setup, openssl has had pkgconfig forever.
[14:48:47 CET] <wm4> maybe it's windows
[14:48:49 CET] <sfan5> cross-compiling, at least that applies for android
[14:48:55 CET] <wm4> or android lolol
[14:49:12 CET] <jkqxz> pkgconfig is fine for cross-compiling, just set the right PKG_CONFIG_PATH.
[14:49:48 CET] <sfan5> i'm not disabling it explicitly
[14:50:11 CET] <sfan5> it is probably looking for arm-linux-androideabi-pkg-config and not finding it
[14:51:14 CET] <kierank> J_Darnley: I would say just push your asm
[14:51:28 CET] <JEEB> yeah, I usually define that in PATH which contains PKG_CONFIG_LIBDIR=/my/android/sysroot/lib/pkgconfig
[14:51:31 CET] <sfan5> ffmpeg breaks with 1.1.0e in this case anyway (errors in libavformat/rtmpdh.c)
[14:51:58 CET] <JEEB> PKG_CONFIG_PATH <= appends to the search path, PKG_CONFIG_LIBDIR <= overrides the search path completely
[14:52:13 CET] <JEEB> thus latter is what you use for cross-compile
[14:52:35 CET] <sfan5> why is this so broken
[14:52:40 CET] <sfan5> i feel like adding libtls support to ffmpeg
[14:52:51 CET] <sfan5> just to avoid the clusterfuck of cross-compiling a working(!) openssl
[14:53:55 CET] <wm4> what's libtls
[14:54:13 CET] <wm4> but yeah, windows and osx got native tls wrappers too, just to avoid openssl and friends
[14:54:49 CET] <sfan5> a tls api designed from the ground up by the libressl guys
[14:54:50 CET] <jkqxz> I imagine it would be welcomed.  Patches for stupid openssl (or openssl not-quite-compatibles like libressl) problems are common.
[14:55:23 CET] <wm4> sfan5: is that the one that required unix file handles, or was that something else?
[14:55:30 CET] <sfan5> dunno
[14:57:40 CET] <durandal_1707> michaelni: your hack to qdrw detecting w/h is just hack
[15:12:29 CET] <durandal_1707> also is dnxhd maintainer still active for ffmpeg?
[15:12:48 CET] <durandal_1707> or just promoting ffmbc ?
[15:18:46 CET] <FishPencil> It looks like FFmpeg only emulates pthreads and not semaphore. How does it handle semaphore synchronization without implementing it?
[15:20:28 CET] <sfan5> the problem with libressl seem to be the "OPENSSL_VERSION_NUMBER >= 0x1010000fL" checks that evaluate to true for libressl
[15:25:10 CET] <iive> durandal_1707: it's quite rare to see michael blocking somebody from doing something. quite the opposite...
[15:28:25 CET] <iive> wm4: michael is having fun coding. He doesn't like getting into shit throwing fights. I find this quite.... logical.
[15:28:25 CET] <wm4> FishPencil: not at all? but emulating semaphores should be simple
[15:28:42 CET] <wm4> iive: then why is he leading the project (and always did, even driving others to fork)
[15:29:11 CET] <wm4> FishPencil: these probably work https://github.com/mpv-player/mpv/blob/master/osdep/win32/pthread.c#L196
[15:29:13 CET] <iive> wm4: because this is coding project, where developers work on code, not on throwing shit on each other.
[15:29:43 CET] <FishPencil> wm4: By replacing sem_wait(); with pthread_cond_wait();?
[15:30:28 CET] <FishPencil> It looks like very few things actually depend on Semaphore
[15:46:41 CET] <cone-439> ffmpeg 03Paul B Mahol 07master:3f878aaedf1e: avcodec/flicvideo: update comment, 24bit support is implemented
[16:03:36 CET] <sfan5> patches go to the mailing list right?
[16:05:38 CET] <JEEB> yes
[16:06:02 CET] <JEEB> ffmpeg-devel at ffmpeg.org
[16:18:40 CET] <sfan5> >Post by non-member to a members-only list
[16:18:45 CET] <sfan5> ugh should've signed up first
[16:23:12 CET] <JEEB> aye
[18:30:26 CET] <BBB> michaelni: regarding that h264_ps, how can an innocent uv_qp_index_offset cause bad behaviour? I guesswhat Im asking is: what is the actual bug (direct invalid behaviour) that this header check fixes?
[18:31:13 CET] <michaelni> BBB, undefined behavior due to integer overflow
[18:31:17 CET] <michaelni> IIRC
[18:31:48 CET] <BBB> oh ok so the value is like near INT_MIN or INT_MAX or so?
[18:31:59 CET] <michaelni> yes i think it was
[18:32:08 CET] <BBB> ok, that makes sense then
[18:32:19 CET] <BBB> (I have no other comments)
[18:32:24 CET] <michaelni> ok
[18:34:29 CET] <BBB> dont be too harsh on wm4s comment btw, there is sometimes a tendency to be too verbose in our errors, causing fuzz files to basically swamp your stderr with useless crap. most errors (fuzz only stuff) dont need wordy errors, just ERROR_INVALIDDATA is good enough. if you feel this is outside the scope of that sort of errors, the string can probably stay (I wouldnt know how common this error is in valid files or developer programming mista
[18:36:12 CET] <wm4> your line cuts off at "mista"
[18:37:14 CET] <BBB> +kes)
[18:37:18 CET] <BBB> stupid irc client :(
[18:38:47 CET] <cone-439> ffmpeg 03James Almer 07master:65d7d46f717c: avcodec/mjpegenc_huffman: add missing header include
[18:38:48 CET] <cone-439> ffmpeg 03James Almer 07master:a267cac92817: avcodec/cbrt_data: add missing header include
[19:22:49 CET] <tmm1> is it possible to select which graphics device to use with QSV?
[19:27:10 CET] <wm4> tmm1: not sure about the current state, but full-fledged control is coming (there are patches in Libav)
[20:53:22 CET] <cone-439> ffmpeg 03Mulvya 07master:8b768f68b054: doc: drawtext options update
[20:56:36 CET] <durandal_1707> can someone kindly review my patches on list?
[21:23:45 CET] <cone-439> ffmpeg 03Paul B Mahol 07master:2c5839254daa: avcodec/dnxhddec: fix decoding of DNxHR HQX 10-bit
[21:25:05 CET] <cone-439> ffmpeg 03Paul B Mahol 07master:fb5c9be82ea8: avcodec/qdrw: add support for 0x0001 code
[22:30:31 CET] <cone-439> ffmpeg 03Michael Niedermayer 07master:5a8fec1b33f2: avcodec/mpegaudiodec_template: Fix multiple runtime error: signed integer overflow
[22:30:32 CET] <cone-439> ffmpeg 03Michael Niedermayer 07master:6871df02d973: avcodec/mpeg4videodec: Check sprite_offset in addition to shifts
[22:30:33 CET] <cone-439> ffmpeg 03Michael Niedermayer 07master:ec849f637e85: avcodec/h264idct_template: Fix several runtime error: signed integer overflow
[22:58:09 CET] <cone-439> ffmpeg 03Paul B Mahol 07master:039011b6b0c2: avcodec: add ScreenPressor decoder
[23:16:24 CET] <durandal_1707> wow Compn is back!
[23:19:06 CET] <Compn> durandal_1707 : sorry i missed your email
[23:19:11 CET] <Compn> it came in with some spam lol
[23:19:29 CET] <durandal_1707> lol
[23:24:57 CET] <durandal_1707> Compn: i think screenflow codec is yet another just zlib codec
[23:29:45 CET] <wbs> RiCON: fwiw, about that rtmpdh patch on the list; have a look at 016387fe0fe3eff1a03ec0673bf4d2967f6cad94 in libav instead
[23:30:20 CET] <wbs> (that patch on the list blindly hardcodes the internal representation of an openssl struct, that now has been made opaque just to allow them to change the layout of it, if needed)
[23:32:08 CET] <kierank> funman: can you delete ffmpeg-sdi
[23:35:20 CET] <kierank> atomnuker: apologies didn't realise there was still a repo in existence
[23:37:21 CET] <sfan5> hm wait it actually does
[23:37:55 CET] <cone-439> ffmpeg 03Marton Balint 07master:436f00b10c06: avcodec/wrapped_avframe: allocate a buffer with padding
[23:41:21 CET] <llogan> atomnuker: what's the reason why "-huffman optimal" was not made default?
[23:45:11 CET] <atomnuker> llogan: I guess the authors are busy and forgot or something
[23:45:38 CET] <atomnuker> should I send a patch or do you have one already?
[23:46:07 CET] <llogan> feel free to do so if you want to. i don't have a patch.
[23:47:22 CET] <atomnuker> patch sent
[23:50:51 CET] <BtbN> that openssl 1.1 fix looks like a horrible patch with potentially bad aftermath if openssl actually changes the struct
[23:51:20 CET] <RiCON> sfan5: ^
[23:51:27 CET] <sfan5> yeah i noticed
[00:00:00 CET] --- Thu Feb 23 2017


More information about the Ffmpeg-devel-irc mailing list