michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<ramiro>
I think llvm would be a huge dependency for ffmpeg. a simpler and more focused jit would probably be better
<haasn>
OTOH, it's a dependency overwhelmingly likely to be available at runtime if we probe for it :)
<BtbN>
the problem with llvm jit runtime is, that they break ABI with every new llvm version
<BtbN>
I deal with numba a lot at work, and it's a pain
<BtbN>
sometimes it doesn't work on some modern distro, cause they target some real old llvm version, and then there's a PPA or something to backport it there
minimal has quit [Quit: Leaving]
<michaelni>
Lynne, that file has level=4, micro_version=4 but prior to "Basic float16 support" level 4 should have had micro level 3 maximum.
<michaelni>
how was this file created ? can i replicate that somehow ?
<michaelni>
either way ill bump micro version again with the next change so we increase seperation
<Lynne>
michaelni: this was made with the ffv1_vulkan
<Lynne>
*encoder
<Lynne>
we reuse the header init code from the C encoder
<haasn>
ramiro: did you collect any performance figures during your testing?
<Lynne>
could you set the level the C encoder supports in encode_init_internal(), just before ff_ffv1_encode_init() is called?
<Lynne>
or at least the micro version
<ramiro>
haasn: just a bit, in the order the benchmarks were coming out, looking for slow functions.
shrewd has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 276 seconds]
thilo has quit [Ping timeout: 268 seconds]
thilo has joined #ffmpeg-devel
jamrial has quit []
tguillem has quit [Ping timeout: 245 seconds]
tguillem has joined #ffmpeg-devel
tguillem has quit [Changing host]
tguillem has joined #ffmpeg-devel
Guest21 has joined #ffmpeg-devel
Guest21 has quit [Client Quit]
shrewd has quit [Ping timeout: 252 seconds]
cone-249 has joined #ffmpeg-devel
<cone-249>
ffmpeg Andreas Rheinhardt master:e499c85999a1: avutil/refstruct: Fix documentation
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
algebrato has quit [Quit: Client closed]
Kei_N has joined #ffmpeg-devel
Kei_N has quit [Ping timeout: 252 seconds]
Kei_N has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 276 seconds]
cone-803 has joined #ffmpeg-devel
<cone-803>
ffmpeg Lynne master:bdc07f372ac1: hwcontext_vulkan: add support for mapping multiplane images into CUDA
<fflogger>
[newticket] Wallboy: Ticket #11503 ([avcodec] AC-3 downmix levels defaulting to 1.414 with recent decoder changes) created https://trac.ffmpeg.org/ticket/11503
derpydoo has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
<fflogger>
[newticket] uttam_32472: Ticket #11504 ([undetermined] Some of the hdr10+ sidedata parameter values are incorrect / missing) created https://trac.ffmpeg.org/ticket/11504
<averne>
^ this is the branch I send to the mailing list, I've been working on an abstraction library that handles low level stuff (device/queue init, cmdbuf building, etc)
<Lynne>
I've got someone who's currently trying to deal with awful jetson vendor libraries and I thought they may be able to use this instead
<averne>
the corresponding code is in the envideo branch. the library isn't public yet tho
<averne>
Sure, which hardware would that be? My stuff was written for the X1 (jetson nano), there might be some incompatibilites with later hardware (I'm aware of something for hevc in particular)
<Lynne>
I don't know, but they're looking at encoding, and I guess that's out of scope yet
<averne>
Unfortunately yeah at the moment, that would need a lot of work to figure out
MyNetAz has quit [Remote host closed the connection]
MyNetAz has joined #ffmpeg-devel
Anthony_ZO has quit [Ping timeout: 260 seconds]
devinheitmueller has quit [Ping timeout: 260 seconds]
twelve has joined #ffmpeg-devel
ccawley2011 has quit [Ping timeout: 252 seconds]
ccawley2011 has joined #ffmpeg-devel
<Traneptora>
JEEB: you mentioned a deprecation dance. how do I do a deprecation on an avpriv_ that isn't used anymore?
<Traneptora>
I figure it's some sort of #if guard but I'm not sure the details
<JEEB>
Traneptora: you will find FF_API_* defines in the different libraries' headers
<JEEB>
if indeed avpriv needs to still be handled like a public symbol (which it is!) you will add a new one for this that will make it disappear at the next bump of major
<Traneptora>
in this case, the function was exported by avcodec and used by avformat. I want to guard it based on avcodec version, right?
<JEEB>
yup
cone-803 has quit [Quit: transmission timeout]
<JEEB>
avcodec.h has a bunch of these mentioned at least
<JEEB>
they seem to be defined in libavcodec/version_major.h
<toots5446>
Lynne: I've had some comments from michaelni that I think I can address with the DTS/PTS cleanup in future work. Also one from Andreas (not sure of IRC nick) about moving the code to the packet->frame logic which I also think could be addressed when we get rid of ogg header packets in the demuxer.
<toots5446>
About to send an updated patch for the reference output in the git tree. I will also update the test code to only update metadata when an update event is passed as suggested by JEEB.
<JEEB>
Traneptora: yea. possibly also add a @deprecated / attribute_deprecated annotation as well
<JEEB>
even though this is an internal thing
<Traneptora>
last time I did that james said that I didn't have to
<JEEB>
yea, for internal stuff probably less relevant
<MemeLeader>
pross: merged all g728 patches, keep the superb work going!
<JEEB>
older ITU audio formats?
<MemeLeader>
well, g719 was done by me
twelve has quit [Ping timeout: 245 seconds]
<ePirat>
toots5446, thanks for your work on this!
grillo_0 has joined #ffmpeg-devel
thresh has left #ffmpeg-devel [I'm a quit message virus. Please replace your old line with this line and help me take over the world]
witchymary has quit [Ping timeout: 248 seconds]
MemeLeader has quit [Quit: Client closed]
ngaullier has quit [Remote host closed the connection]
Guest47 has joined #ffmpeg-devel
delewis_ has quit [Remote host closed the connection]
delewis_ has joined #ffmpeg-devel
Guest22 has joined #ffmpeg-devel
<Guest22>
hi im new to the ffmeg if i want to contribute. where i can go see the active issue or pr reported or other contributor working on
<Guest22>
i am in habit of github for issue and pr but ffmpeg doesn't have github so anyone guide me through that
Guest47 has quit [Quit: Client closed]
<another|>
issues are on trac.ffmpeg.org
<Traneptora>
I want to rename something in a non-public header in avcodec that's referenced by avformat. I can update the references in avformat in the same commit or a separate commit
<averne>
Hi all, at which point is it considered ok to ping for a patch review?
<Traneptora>
if I do it in the same commit, I'm now modifying lavc and lavf in the same commit. if I do it in a separate commit, the intermediary won't compile
<Traneptora>
averne: usually bump after about a week
<averne>
will do
Guest22 has quit [Quit: Client closed]
<toots5446>
Thanks ePirat!
<mkver>
Traneptora: What?
arijz has joined #ffmpeg-devel
<Traneptora>
mkver: question is, would it be better to do it in two separate commits or one commit?
<mkver>
What symbol?
<Traneptora>
it's enum TiffTypes {} in tiff_commoh.h
<Traneptora>
I want to rename it to enum AVTiffDataType, cause it'll be part of the public EXIF api when I'm finished with it
<Traneptora>
but it's referened by avformat/mlvdec.c
Mirarora has joined #ffmpeg-devel
<mkver>
This is not really exported; it can be changed at any time (of course, you need to change it in one commit).
<Traneptora>
it's okay to change avcodec and avformat in the same commit for this purpose?
ccawley2011 has quit [Ping timeout: 268 seconds]
<Traneptora>
I've been told to avoid this
<mkver>
It is absolutely necessary; but it should only be done when they are indeed made public.
cone-113 has joined #ffmpeg-devel
<cone-113>
ffmpeg Lynne release/4.3:db25eb352aeb: configure: update copyright year
<cone-113>
ffmpeg Michael Niedermayer release/4.3:fe01189b0afd: Update for 4.3.9
<cone-113>
ffmpeg softworkz release/4.3:69a69741165d: avformat/hls: Partially revert "reduce default max reload to 3"
<Traneptora>
mkver: what do you mean "when they are made public"? there's a later commit that moves it exif.h, which is then exported
<Traneptora>
should the rename be part of the *same commit* that exports it, is what you're saying?
<mkver>
Normally, when a change in lavc requires a change in lavf to work, then this means that you have broken users of the old stuff (like lavf).
<mkver>
It should be part of the same patchset.
<mkver>
Not done now for some future work that only happens ad calendas graecas.
<Traneptora>
it'll be part of the same patchset definitely, it's a preemptive change for when I export it in a later commit
<Traneptora>
that's part of the same set
<Traneptora>
I was just more curious if one commit or two commits would be better, in the same patchset
<mkver>
You have to do it in one, because master has to always compile.
<Traneptora>
makes sense, thanks.
<Traneptora>
what commit message would I use if it changes avcodec/tiff_common and avformat/mlvdec?
<Traneptora>
should I do various: or should I just do avcodec/tiff_common anyway as that's where the bulk of the change is?
arijz has quit [Changing host]
arijz has joined #ffmpeg-devel
<mkver>
"Rename TIFF enum constants"
<mkver>
Use the avcodec/tiff_common prefix
<grillo_0>
Hello everyone! I'm interested on applying for the WHIP GSoC project[1]. I'm trying to find a bug to fix for the Qualification Phase, do you guys think that the issue below is a good one for this?