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 7.0.2 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
thilo_ has quit [Ping timeout: 272 seconds]
thilo_ has joined #ffmpeg-devel
thilo_ has quit [Changing host]
thilo_ has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
haihao has quit [Ping timeout: 272 seconds]
haihao has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
Warcop has joined #ffmpeg-devel
NotWarcop has quit [Ping timeout: 260 seconds]
cone-063 has joined #ffmpeg-devel
<cone-063>
ffmpeg James Almer master:0dd6f1d5bf12: avformat/mov: fix setting index of LCEVC enhancement stream
System_Error has quit [Ping timeout: 260 seconds]
mkver has joined #ffmpeg-devel
jamrial has quit []
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 245 seconds]
arbitercoin has joined #ffmpeg-devel
arch1t3cht5 has joined #ffmpeg-devel
<cone-063>
ffmpeg Zhao Zhili master:e18b46d95fad: swscale/aarch64: Fix rgb24toyv12 only works with aligned width
arch1t3cht has quit [Ping timeout: 248 seconds]
arch1t3cht5 is now known as arch1t3cht
Kei_N_ has joined #ffmpeg-devel
Kei_N has quit [Ping timeout: 245 seconds]
pross has quit [Ping timeout: 265 seconds]
mkver has quit [Ping timeout: 260 seconds]
haihao has quit [Ping timeout: 276 seconds]
MrZeus has joined #ffmpeg-devel
haihao_ has joined #ffmpeg-devel
cone-063 has quit [Quit: transmission timeout]
MrZeus has quit [Ping timeout: 260 seconds]
microchip__ has joined #ffmpeg-devel
elChippo has joined #ffmpeg-devel
microchip__ has quit [Client Quit]
microchip_ has quit [Ping timeout: 265 seconds]
elChippo is now known as microchip_
wyatt8740 has quit [Remote host closed the connection]
mkver has joined #ffmpeg-devel
wyatt8740 has joined #ffmpeg-devel
wyatt8750 has joined #ffmpeg-devel
wyatt8740 has quit [Ping timeout: 252 seconds]
ngaullier has joined #ffmpeg-devel
ngaullie has joined #ffmpeg-devel
ngaullier has quit [Ping timeout: 244 seconds]
Warcop has quit [Ping timeout: 264 seconds]
rtwo has quit [Remote host closed the connection]
mark4o has joined #ffmpeg-devel
markh has quit [Ping timeout: 260 seconds]
mark4o is now known as markh
ccawley2011 has joined #ffmpeg-devel
<wbs>
FYI, fate is broken with MSVC, patch on the ML
<elenril>
wbs: what is msvc-specific about that?
<wbs>
elenril: nothing really, just that MSVC probably is the main case that hits the fallback, non-gcc, non-clang
<elenril>
actually on my debian the header is provided by glibc, not compiler
<wbs>
right
<elenril>
but anyway, I still don't see what exactly is broken
<elenril>
the upper bits should all be zero, no?
<wbs>
yes
<wbs>
but we're shifting to the _left_
<wbs>
counting the number of bits we need to shift until value == 0
<wbs>
but we initialize zeros to the size of the input data
<elenril>
oh, right
<elenril>
> unsigned long long mask = (1ULL << (size * CHAR_BIT)) - 1;
<elenril>
isn't that an invalid shift when size == sizeof(unsigned long long)?
<wbs>
I was pondering that... yes, it unfortunately probably is
<elenril>
I got hit by this a few times before
<elenril>
it's really annoying how there's no "clean" way to make a bitmask of n bits
<wbs>
we could also do ((1ULL << (size/2 * CHAR_BIT) << (size/2*CHAR_BIT)). not very pretty :P
<elenril>
would that even help?
<wbs>
yes, it's valid to shift bits out to the left side, but you can't do one single shift wider (or equal to) the size of the value
<elenril>
it wouldn't be UB, but would the result be right?
<wbs>
as for the total expression, I lost the final -1 at the end
<elenril>
I mean you still shift your 1 out of the range of the type
<elenril>
oh right, -1 is the correct answer
<wbs>
yes, but that's valid to do for unsigned types
<wbs>
exactly, it's just the single shift that is too wide that is the issue
<elenril>
I wonder why is it not shifting to the right instead
microlappy has joined #ffmpeg-devel
microlappy has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<wbs>
elenril: shifting to the right requires a condition for zero and ends up marginally less elegant. but correct also is better than elegant but wrong
<elenril>
when taking your patch into consideration, shifting to the right seems cleaner and easier to understand to me
<wbs>
yep
<wbs>
that's why I just sent a patch that does that :P
<elenril>
I could swear I saw exactly the same code in my editor 30 minutes ago
<elenril>
are you a wizard?
<wbs>
no, but there are only so many ways to implement this :P
___nick___ has joined #ffmpeg-devel
<Daemon404>
why didnt we just remove r_frame_rate again?
<elenril>
it might be useful for remuxing mpeg4 from nut to avi?
<JEEB>
o_O
<Daemon404>
my sarcasm meter just exploded
<wbs>
elenril: fyi, as reported yesterday as well, the fate-hevc-mv-switch test seems to be failing on almost everything that isn't x86_64
cone-727 has joined #ffmpeg-devel
<cone-727>
ffmpeg Martin Storsjö master:b4d9fa6cb93a: compat: Fix the fallback definition of stdc_trailing_zeros
<elenril>
wbs: I'm on it, but neither usan nor asan nor valgrind show anything here
<elenril>
guess I'll try a 32bit build
<wbs>
that's probably the easiest, yeah
<elenril>
nope, no luck
<elenril>
forcing the compat stdbit.h doesn't trigger it either :/
<wbs>
do you have wine available? it triggers with llvm-mingw in i686 mode at least, presumably also with MSVC and other compilers
<elenril>
I have a setup with msvc in wine, based on your scripts
<elenril>
but I've never tried debugging in it
<wbs>
printf debugging works the same everywhere :P
<wbs>
oh, wait, I was mistaken. there was another failure in the llvm-mingw i686 job, a stray checkasm failure
<wbs>
so it's possible that the hevc issue only appears on other architectures
elenril has quit [Ping timeout: 252 seconds]
elenril has joined #ffmpeg-devel
elenril has quit [Ping timeout: 276 seconds]
<kurosu>
wbs: not exactly, under Windows, you often need twice as many printfs to fix race conditions
rvalue has quit [Ping timeout: 252 seconds]
rvalue has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
<Daemon404>
kurosu, the best way to fix race conditions on windows is to port your program to linux
<JEEB>
fun, decided to test 8875da347a and apparently now streams that don't have inputs may cause stats to think things aren't moving fast enough :D
<JEEB>
(specifically, subtitles)
<JEEB>
I just have warnings on my test stuff that check whether the stats say it's going under 1.0 by too much, and getting a bunch of them now :) will have to check if I'm getting it with current master as well
Krowl has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
<kurosu>
Daemon404: and a RT kernel (is that still a thing?)
<Daemon404>
no clue
<jdek>
courmisch: lpads added (I feel like this should just be in the func macro?) and 128 bench added
cone-727 has quit [Quit: transmission timeout]
haihao_ has quit [Ping timeout: 246 seconds]
haihao has joined #ffmpeg-devel
cone-627 has joined #ffmpeg-devel
<cone-627>
ffmpeg Michael Niedermayer master:fde8637fda8e: avformat/asf: Check picsize
<cone-627>
ffmpeg Michael Niedermayer master:f2aba7bd9f26: MAINTAINERS: Fix typo in web repository
<cone-627>
ffmpeg Michael Niedermayer master:ad35eaf848bb: avcodec/osq: Treat sum = 0 as k = 0
<cone-627>
ffmpeg Michael Niedermayer master:9d25b9665edb: avcodec/cbs_h266_syntax_template: Check bit depth with range extension
<cone-627>
ffmpeg Michael Niedermayer master:e9f588af9530: avcodec/osq: use unsigned for decorrelation
<cone-627>
ffmpeg Michael Niedermayer master:f27c8b04d305: avcodec/jfdctint_template: use unsigned z* in row_fdct()
<cone-627>
ffmpeg Michael Niedermayer master:36924fa30627: avcodec/aac/aacdec: use correct index in deallocation
<cone-627>
ffmpeg Michael Niedermayer master:74385dd496bd: avcodec/encode: Check bitrate
<cone-627>
ffmpeg Michael Niedermayer master:9a823fbcfae3: tools/target_dec_fuzzer: Adapt threshold for RASC
<cone-627>
ffmpeg Michael Niedermayer master:c3a1cbbf5d99: avcodec/eacmv: Check input size for intra frames
<cone-627>
ffmpeg Michael Niedermayer master:5ded752241ae: tools/target_dec_fuzzer: Add threshold for SRGC
<cone-627>
ffmpeg Michael Niedermayer master:6190e826471f: tools/target_swr_fuzzer: Limit the number of samples
<cone-627>
ffmpeg Michael Niedermayer master:050b5e85cbe6: avcodec/svq3: Check for minimum size input
<cone-627>
ffmpeg Michael Niedermayer master:8367d7e18456: avcodec/imm4: Check input size
<cone-627>
ffmpeg Michael Niedermayer master:5f471b500c4f: avcodec/sgirledec: Check input length
<cone-627>
ffmpeg Michael Niedermayer master:03050a0d9092: avcodec/vble: Allocate buffer later
<Daemon404>
interesting that the 'open sourced' winamp code contains a ton of binaries and code from proprietary places
<Daemon404>
i suspect they werent supposed to do that
<Daemon404>
(dolby, fraunhofer, duck/on2)
<jdarnley>
That's out now? I gotta grab it
<Daemon404>
before it's pulled due to angry rights people? ;)
<JEEB>
reminds me of how symbian open source got pulled
<jdarnley>
I don't care about any of the multimedia crap (#ffmpeg-devel gasps in horror) I just want the library database code
<JEEB>
lol, I love the ATLMFC fix note in readme :D
<King_DuckZ>
hey all, I'm working on some bug that seems to be in mpv, it seems to be caused by a combination of invalid udp data causing mpv to pause and then not detecting the end of stream afterwards, while the udp buffer in ffmpeg fills up hopelessly
<King_DuckZ>
I can now manage the paused state somewhat, but it'd be useful if I could flush that udp buffer from mpv, is there an API for that, or should I tear down all the codecs and re-init everything?
elenril has joined #ffmpeg-devel
<jdarnley>
I see it posted on hacker news and I'm going to guess that TOS enforcers are going to get it removed before any rights holders
<dukesook>
Hello, I'm looking at adding uncompressed MP4s to FFMPEG according to ISO/IEC 23001-17:2024. Could anyone give me some guidance on completing this task?
<King_DuckZ>
haha it's got 32 forks already xD
<cone-627>
ffmpeg Michael Niedermayer master:10d8b6f17c8c: doc/APIchanges: update
<cone-627>
ffmpeg Michael Niedermayer master:27c6f5f6d2f1: Changelog: Mark for 7.1
<cone-627>
ffmpeg Michael Niedermayer master:e1094ac45d3b: */version.h: bump minor versions for release/7.1
<cone-627>
ffmpeg Michael Niedermayer master:38e224c2ba09: */version.h: bump after release/7.1 branch
<cone-627>
ffmpeg Michael Niedermayer master:2af5e7f50da4: Changelog: Add version <next>
<cone-627>
ffmpeg Michael Niedermayer release/7.1:HEAD: Changelog: Add version <next>
Krowl has joined #ffmpeg-devel
<cone-627>
ffmpeg Michael Niedermayer master:f7d709fed892: doc/APIchanges: Add 7.1 Cut line
<jamrial>
michaelni: fyi, there's an outstanding issue in master (and also in release/7.1), which needs to be fixed before tagging
<michaelni>
jamrial, what issue ? will you fix it :) ?
<jamrial>
michaelni: fate-hevc-mv-switch fails on some targets, and elenril is looking at it
<jamrial>
does the source code show how it kicks the llama's ass?
<King_DuckZ>
I haven't checked yet, but in case they decide to pull it from github
blb_ has quit [Ping timeout: 276 seconds]
blb has joined #ffmpeg-devel
<Daemon404>
are we still doing the silly thing where we branch a release off master at release time?
<Daemon404>
rather than branch it, freeze that branch except for bug fixes for some time, then release?
<another|>
do people still actually care about winamp?
<Daemon404>
another|, nostalgia is a strong thing
<another|>
hmmkay
<another|>
never really used it. have no nostalgia
<King_DuckZ>
I don't, it's just for those who do care, I like Strawberry personally
<Daemon404>
i dont even play music in my computer anymore
<Daemon404>
s/in/on/
<Daemon404>
mostly because dell has terrible audio in its laptops.
<King_DuckZ>
from a quick look I don't think a Linux port is something I'd attempt, so my interest just fell from 2 to -50 :|
<elenril>
Daemon404: the above is a branch for the release
<Daemon404>
elenril, ah ok
<elenril>
that should only get bugfixes from now on
<another|>
I had an mpd before moving
<michaelni>
jamrial, fate-hevc-mv-switch is triggered by cpuflags 0, anyway will leave that to elenril, i dont know the code ...
<Daemon404>
elenril, i was wondering how mv-hevc got in the release branch
<Daemon404>
since it only just went in
<Daemon404>
and 7.1 is out really soon supposedly.
<elenril>
I pushed it yesterday
<elenril>
the branch happened 5 minutes ago
<Daemon404>
cool... under a day before release
<King_DuckZ>
btw if anyone can help with my original question please...
<Daemon404>
not really what i meant
<Daemon404>
more like branch 1-2 *weeks* before
<Daemon404>
a freeze.
<elenril>
like a pre-freeze freeze?
<another|>
a fridge?
<Daemon404>
... i dont think you really understand code freezes
<Daemon404>
if you think it lasts a day or two
<Daemon404>
before release
dukesook has quit [Quit: Client closed]
<elenril>
who is saying the release is 2 day away?
<Daemon404>
> I intend to make the 7.1 release from that branch within a week!
<Daemon404>
just now on the ml
<elenril>
that....indeed seems a bit hasty
<King_DuckZ>
BUILD_BUG_ON(offsetof(Slice, pos) != 15764); my interest is in free fall at this point
<michaelni>
fedora asked us a month ago if we can get the release done around october 1st as Fedora 41 final freeze is October 15. So it would be nice if we can get 7.1 released before that
vipyne has joined #ffmpeg-devel
<michaelni>
well, not litterally "fedora" but you know what i meant :)
<wbs>
elenril: managed to repro the fate issue yet? did you see michaelni's comment that cpuflags 0 triggers it?
<elenril>
yes, that does seem to trigger it
<elenril>
looking at it atm
<elenril>
I've also had a little power outage today, so it's not progressing quickly
<Daemon404>
isnt winamp owned by some crypto bros nowadays?
<Daemon404>
or am i thinking of something else
<Daemon404>
if so, i am not surprised it's weird AF
<j-b>
good morning sarcasm
<elenril>
huh, could it be just that I'm missing bitexact sws flags
<JEEB>
reminds me of my first experience in 2013 or so
<Daemon404>
> On December 6, 2022, Winamp 5.9.1 was released, adding a music NFT playback feature. Users are able to add music NFTs on Ethereum and Polygon to the media library by connecting to the Metamask wallet.
<Daemon404>
yep...
<elenril>
wbs: can you test the patch I just sent on some weird architecture?
<King_DuckZ>
oh great, a non-free windows-only player that understands shitcoins is exactly what I needed!
<Daemon404>
this dump is for the old codebase at least
<wbs>
elenril: LGTM, thanks!
<jamrial>
elenril: lol, happened to me too
<elenril>
I was getting afraid it's a much deeper bug
<cone-627>
ffmpeg Anton Khirnov master:9dc0edd4546b: tests/fate/hevc: use bitexact scaling flags for fate-hevc-mv-switch
<cone-627>
ffmpeg Anton Khirnov release/7.1:f459c56b0fe3: tests/fate/hevc: use bitexact scaling flags for fate-hevc-mv-switch
<cone-627>
ffmpeg sunyuechi master:ba7d0d5fc391: lavc/vvc_mc: R-V V avg w_avg
<cone-627>
ffmpeg Rémi Denis-Courmont master:4936bb25083e: lavc/h264dsp: optimise R-V V weight for shorter heights
<cone-627>
ffmpeg Rémi Denis-Courmont master:459a1512f13a: lavc/h264dsp: unroll R-V V weight16
<cone-627>
ffmpeg Rémi Denis-Courmont master:6611bf548497: lavc/h264dsp: optimise R-V V biweight for shorter heights
ngaullie has quit [Ping timeout: 244 seconds]
<Marth64>
ahhh Winamp
<Marth64>
blast from the past
<courmisch>
sounds like what a Napster pirate would say
<courmisch>
also anyone who findg bugs in my code will reported to the Truth Protection Agency
Guest8 has joined #ffmpeg-devel
Guest8 has quit [Quit: Client closed]
Krowl has quit [Read error: Connection reset by peer]
<cone-627>
ffmpeg James Almer master:853c16538614: avcodec/cbs_h265: fix valid range for {left,right}_view_id
<cone-627>
ffmpeg James Almer master:b8e74aa1cb71: avcodec/cbs_h265: num_ref_displays can be up to 32
<cone-627>
ffmpeg James Almer master:feef6928817d: avcodec/hevc/sei: num_ref_displays can be up to 32
<cone-627>
ffmpeg James Almer release/7.1:00ce36d037b2: avcodec/cbs_h265: fix valid range for {left,right}_view_id
<cone-627>
ffmpeg James Almer release/7.1:a6a346e15444: avcodec/cbs_h265: num_ref_displays can be up to 32
<cone-627>
ffmpeg James Almer release/7.1:2db706374e54: avcodec/hevc/sei: num_ref_displays can be up to 32