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 6.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
mkver has quit [Ping timeout: 264 seconds]
HarshK23 has quit [Quit: Connection closed for inactivity]
kurosu has quit [Quit: Connection closed for inactivity]
averne has quit [Ping timeout: 255 seconds]
Marth64 has quit [Ping timeout: 264 seconds]
Marth64 has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
averne has joined #ffmpeg-devel
<compn>
if i may make a suggestion to the smart people out there
<compn>
when you get a mail on the list and it is like "what is going on, why would you do this, who made this decision and where"
<compn>
take a minute and breathe
<compn>
then ask as a reply "what would you like me to do?"
<compn>
it will solve about 99% of problems with people in this project
<compn>
now i highlight people Daemon404 kierank BBB elenril courmisch jamrial haasn wbs jkqxz devinheitmueller
<compn>
and also michaelni :D
thilo has quit [Ping timeout: 246 seconds]
* compn
looks at j-b
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
<compn>
michaelni, other people can get upset about questions. so when you ask "where did this discussion take place" , instead, you could rephrase your question. example: "i would like to join in on ffmpeg chatting if there is somewhere new besides irc/ml. please give me the details."
<compn>
this has been a public service announcement from your neighborhood compn. see ya next week!
sdc has joined #ffmpeg-devel
MrZeus has quit [Ping timeout: 264 seconds]
<Marth64>
good day
Traneptora has joined #ffmpeg-devel
sdc has left #ffmpeg-devel [#ffmpeg-devel]
cone-831 has quit [Quit: transmission timeout]
Marth64 has quit [Remote host closed the connection]
rajivharlalka has quit [Quit: Connection closed for inactivity]
derpydoo has quit [Ping timeout: 256 seconds]
Marth64 has joined #ffmpeg-devel
cone-177 has joined #ffmpeg-devel
<cone-177>
ffmpeg Nuo Mi master:1e174120d48a: avcodec/vvcdec: frame_context_setup, set fc->ref to NULL
<Traneptora>
did we end up breaking the ABI?
<Traneptora>
bumping*
<Lynne>
not yet
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 264 seconds]
AbleBacon has quit [Read error: Connection reset by peer]
Marth64 has quit [Ping timeout: 252 seconds]
Marth64 has joined #ffmpeg-devel
qeed has quit [Remote host closed the connection]
averne has quit [Ping timeout: 264 seconds]
MetaNova has quit [Remote host closed the connection]
MetaNova has joined #ffmpeg-devel
averne has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 264 seconds]
Marth64 has joined #ffmpeg-devel
averne has quit [Ping timeout: 272 seconds]
averne has joined #ffmpeg-devel
Teukka has quit [Read error: Connection reset by peer]
HarshK23 has joined #ffmpeg-devel
cone-177 has quit [Quit: transmission timeout]
Teukka has joined #ffmpeg-devel
Teukka has quit [Changing host]
Teukka has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 255 seconds]
Marth64 has joined #ffmpeg-devel
wellsakus has quit [Remote host closed the connection]
epony has quit [Remote host closed the connection]
Krowl has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
<kierank>
Could you please send a new one to the patchwork?
<kierank>
What does it mean to send a new patch to patchwork
<elenril>
it's like pushing to our github
<kierank>
How does one send to patchwork
<kierank>
It didn't pick up my original
<kierank>
Doesn't see any of mine, I guess because I use an attachment
ccawley2011 has joined #ffmpeg-devel
epony has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 264 seconds]
Marth64 has joined #ffmpeg-devel
kurosu has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
ccawley2011 has quit [Read error: Connection reset by peer]
<Daemon404>
you do it by moving to gitlab
<JEEB>
^
<JEEB>
kierank: I think patchwork does read some attachments but has a filter on MIME type, so if your email client sends the "wrong" one it will not pick it up
<Daemon404>
im unclear why it needs to be in patchwork
<Daemon404>
to kick of fate?
<Daemon404>
off*
<JEEB>
yea
<Daemon404>
something something gitlab ci :>
Marth64 has quit [Ping timeout: 255 seconds]
<JEEB>
yea
Marth64 has joined #ffmpeg-devel
<elenril>
mumble mumble gitlab again
<elenril>
it's getting boring
<elenril>
you can talk about it forever, nothing is going to happen until someone actually starts working on it
<JEEB>
indeed
<JEEB>
the problem is if we go by the plan where we self-host first until we figure out that better to share videolan infra? :P
<JEEB>
or do we just share infra with videolan
<elenril>
no, that's not the problem
<elenril>
the problem is, there needs to be a person who pushes this forward
<elenril>
all the other problems are resolvable
<JEEB>
it's not the problem but it does change how much work that person pushing it forward needs to do :P
<elenril>
a lot
<elenril>
it's a lot of work either way
<Daemon404>
presumably we have to start with the basics like who hosts the he anf where
<Daemon404>
etc
<elenril>
which is why nobody is volunteering
<elenril>
Daemon404: I'd say we need to start before that
<elenril>
with things like, why are we switching to something at all
<elenril>
what the advantages and disadvantages are
<Daemon404>
i think tgats mostly a you problem at this point.
<Daemon404>
noone else
<JEEB>
visibility and status of merge requests. easier CI integration.
<nevcairiel>
visibility/tracking of patches/MRs would be the prime reason for me, a ML is terrible for that, and patchwork isnt really any better
<BBB>
elenril: you don't have to use gitlab, github has similar functionality 8)
<elenril>
sigh
<JEEB>
yes, you just want someone to write some things down on the ML, right?
<BBB>
compn: thanks for the service announcement. But I still don't know what to do now
<nevcairiel>
I can already see how its going to go "GitLab (or competing solutions) allow us to more easily track patches and offer visibility in open MRs" - response: "I can do that fine now, maybe get more organized" :P
<BBB>
I can't click a patch(set) and auto-merge it; I can't leave comments attached to a code chunk and see revisions to that code chunk and how they address my comment in one integrated place
<JEEB>
elenril: and btw on your side I'd request you to check the current version of gitlab cli tools and make a list of required features that you need
<JEEB>
so that we can add those as prerequisites and possibly pay people to develop them
<JEEB>
nevcairiel: if we don't want to go straight to TC, we do a GA vote on "Would a system like Github/Gitlab help you with seeing the state of merge requests?" (yes|no)
<Lynne>
leaving github in there is a sure way of getting a no
<mkver>
elenril: Do you plan on opposing Gyan's s302m patch?
<JEEB>
sure, it's there just as an example
<Lynne>
I thought some were undecided on whether to go for gitlab or gittea
<Lynne>
and wanted a discussion
<elenril>
mkver: did he ping it or something?
<JEEB>
jdek tested the cli tools for gitea
<JEEB>
they seemed to not really work well for him
<elenril>
i wanted to look more closely at the details
<mkver>
Yesterday he announced to push it in three days (i.e. the day after tomorrow).
<elenril>
grmbl
jamrial has joined #ffmpeg-devel
<elenril>
thank's for telling me anyway
<JEEB>
was that the set with a decoder or so which you can force on a stream?
<elenril>
that was the set with a decoder that looks like a demuxer
<JEEB>
since one of the things I have on my Eternal Backlog needs bitstream reading from supposed PCM streams
<JEEB>
huh, can a decoder output a different codec_id ? or this calls the decoder internally?
<JEEB>
(thus maybe it should be a bsf?)
<nevcairiel>
blame weird formats that pack things into things while being in things
<JEEB>
esp. since we now can have bsfs at demux
<JEEB>
nevcairiel: yup
<nevcairiel>
stupid pro uses that refuse to replace their stuff and just want to shuffle it through existing infra
<JEEB>
in my case it's old capture cards that were used for TV capture before decryption of .jp TV got easily available
<nevcairiel>
maybe next time design forward looking :P
<JEEB>
the receivers would output the AAC bitstreaming :D
<Daemon404>
things in things in things - is this iamf?
<jamrial>
<@nevcairiel> blame weird formats that pack things into things while being in things <- iamf in mp4... i'm suffering this right now
<Daemon404>
... yep
<Daemon404>
:D
<jamrial>
lol
<JEEB>
:D
<elenril>
nevcairiel: i still don't see why it can't be handled inside the demuxer
<elenril>
but then i didn't look at it closely enough
Krowl has quit [Read error: Connection reset by peer]
<nevcairiel>
it might
Marth64 has quit [Ping timeout: 272 seconds]
Marth64 has joined #ffmpeg-devel
<Lynne>
<durandal_1707> share this with elenril/devel channel: with -fsanitize=thread and filters that use ff_inlink_consume_samples() it gives errors for threads=X, for X > 1; this may give wrong output under rare cases, see doom9 forum thread about arnndn filter usage (where threads=1 is workaround)
<JEEB>
( ^_^)b
MrZeus has joined #ffmpeg-devel
Gramner has quit [Remote host closed the connection]
<BBB>
sounds like gstreamer
<BBB>
things in things in things in things
<BBB>
the codec outputing another codec was an explicit design goal of gstreamer
<BBB>
(not saying it's good or not, plz don't ban me)
<Lynne>
our grand unification theory: avpackets == avframes
<BBB>
now all we need is a free-form string to describe muxing format or codec id
drv has quit [Quit: No Ping reply in 180 seconds.]
<BBB>
and a list of key/value pairs of properties like channels or width or subtitle type
Krowl has joined #ffmpeg-devel
<BBB>
and we're all gstreamer
<BBB>
\o/
* BBB
runs
MrZeus_ has joined #ffmpeg-devel
<bencoh>
that free-form key/value thing isn't that bad in the end tbf, it's quite flexible and not so expensive (compared to video processing)
<elenril>
/stab BBB
<elenril>
bencoh: that flexibility is a curse IMO
<elenril>
it means all your APIs are now vague and underdefined
MrZeus has quit [Ping timeout: 256 seconds]
<Daemon404>
replace everything with void* like enlightenment
<JEEB>
spank spank naughty programmer
<Daemon404>
that is the one
navi has joined #ffmpeg-devel
<Lynne>
would make it considerably less annoying to use avbufferrefs for structs, no need to cast uint8_ts
drv has joined #ffmpeg-devel
<elenril>
you should be using refstruct for structs these days
cone-885 has joined #ffmpeg-devel
<cone-885>
ffmpeg Andreas Rheinhardt master:964693324a75: avformat/avformat: Remove reference to removed setter
<cone-885>
ffmpeg Andreas Rheinhardt master:b84d286ac704: avfilter/vf_(bwdif|yadif)_cuda: Remove unused variables
<cone-885>
ffmpeg Andreas Rheinhardt master:8a2c8687bdb0: avfilter/af_hdcd: Drop a redundant log
<Lynne>
refstruct is lavc-only
<elenril>
that was supposed to be changed eventually
Krowl has quit [Quit: Krowl]
Gramner has joined #ffmpeg-devel
psykose has joined #ffmpeg-devel
drv has quit [Quit: No Ping reply in 180 seconds.]
drv has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 260 seconds]
Marth64 has joined #ffmpeg-devel
<kierank>
12:01:30 <• JEEB> kierank: I think patchwork does read some attachments but has a filter on MIME type, so if your email client sends the "wrong" one it will not pick it up
<kierank>
How does one set mimetype in Gmail
<JEEB>
ex-actly
j-b has quit [Changing host]
j-b has joined #ffmpeg-devel
jamrial has quit [Ping timeout: 255 seconds]
sgm has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 268 seconds]
Marth64 has joined #ffmpeg-devel
<cone-885>
ffmpeg Anton Khirnov master:1e7d2007c3ac: all: use designated initializers for AVOption.unit
<Marth64>
i can help with GitLab migration. i can't host though.
<Marth64>
i have done it before
<Marth64>
granted a few years back though
<JEEB>
yeh, I think at the end of the day we will not end up hosting it ourselves, duplicating the work done by videolan.
jarthur has joined #ffmpeg-devel
<Marth64>
can't blame you. pong me if you need ETL scripts etc
<Marth64>
s/pong/ping
<cone-885>
ffmpeg Anton Khirnov master:86417b759f78: lavfi: get rid of FF_INTERNAL_FIELDS
tufei_ has joined #ffmpeg-devel
tufei has quit [Ping timeout: 255 seconds]
Marth64 has quit [Ping timeout: 260 seconds]
Marth64 has joined #ffmpeg-devel
AbleBacon has joined #ffmpeg-devel
epony has quit [Remote host closed the connection]
epony has joined #ffmpeg-devel
ccawley2011 has joined #ffmpeg-devel
<Lynne>
I just realized, we're dropping the old channel layout API
<Lynne>
that will break a lot of users, which is fine, enough time has passed
<Lynne>
but we should take a look at resetting the micro version to 0 since we're breaking everyone anyway
ccawley2011 has quit [Read error: Connection reset by peer]
<Daemon404>
easiest way to contact paul is in the comments on kostya's blog
epony has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 255 seconds]
Marth64 has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
MrZeus__ has joined #ffmpeg-devel
MrZeus_ has quit [Ping timeout: 264 seconds]
<kierank>
lol
<j-b>
I understand that elenril is recoding GStreamer inside FFmpeg
Marth64 has quit [Ping timeout: 264 seconds]
Marth64 has joined #ffmpeg-devel
<Daemon404>
montreal sounding like a better idea
<JEEB>
:F
<JEEB>
:D
<j-b>
"everything is a filter"
<j-b>
"everything is an option"
<j-b>
:D
<j-b>
<3 elenril
<Daemon404>
have your company pay $500 to get access to elenril
<kierank>
if you want demux inside demux
<kierank>
it'll be inevitable
<Daemon404>
kierank, already there for playlist demuxers
<kierank>
#upipediditfirst
<BBB>
ggmpeg?
<Daemon404>
i expect a full dlopen based plugin system
<kierank>
Daemon404: no need, embedded companies just ship an ffmpeg 4.4 source tree
<kierank>
and their blobs inside
<Daemon404>
seems too new
Marth64 has quit [Ping timeout: 252 seconds]
Marth64 has joined #ffmpeg-devel
<Marth64>
hello
<Lynne>
kierank: nah, they give you patches to apply and binaries
<Lynne>
at least that's what that gpu j2k decoder did
<kierank>
Haha they don't give you patches most of the time
<kierank>
Just a tarball
<compn>
j2k lolol
qeed has joined #ffmpeg-devel
<compn>
surprised how fast ffv1 has replaced jpg2000
epony has quit [Remote host closed the connection]
Marth64 has quit [Ping timeout: 268 seconds]
Marth64 has joined #ffmpeg-devel
<thardin>
oh yeah that russian guy
<thardin>
we couldn't get his thing to run
<JEEB>
Lynne: btw did you already have a limited patch around which can be applied & backported for the AV1 vulkan hwaccel? since you thought your ongoing patch set was too large to backport
<pal>
>surprised how fast ffv1 has replaced jpg2000< ---> I have not seen such evidence other that in specific domains
<thardin>
it also doesn't really matter since every standard has to be supported indefinitely
<JEEB>
or should I attempt to poke at a configure patch that disables av1 vulkan hwaccel if the official header is found?
<thardin>
in practice.
<Lynne>
JEEB: can't replicate the build fail here
<Lynne>
could you check if renaming works?
<JEEB>
I will check the mpv CI build logs for the exact failure. then will try to repro it locally
<JEEB>
yea it looks like the system vulkan header gets found
<JEEB>
or do you mean just s/MESA/KHR/?
<JEEB>
and of course the CI clone log doesn't actually log the hash (*stab*)
Marth64 has quit [Ping timeout: 264 seconds]
Marth64 has joined #ffmpeg-devel
<Lynne>
JEEB: I mean header rename
<JEEB>
roger
<j-b>
elenril: lol
Marth64 has quit [Ping timeout: 246 seconds]
Marth64 has joined #ffmpeg-devel
<JEEB>
ok, repro'd locally with 1.3.277
<Lynne>
JEEB: oh... I was using 255
<Lynne>
that figures
<JEEB>
I love how they just forgot to update one of their build systems for that tag, so main has +1 commit which fixes that
mkver has quit [Ping timeout: 252 seconds]
<JEEB>
also I love how the includes in libavcodec/vulkan_decode.h are all "", so in theory the local version should trump. unless the official header gets indirectly included through another included system header :)
<courmisch>
source tarball? wow, pretty decent. Of late I mostly see source code *and* build artifacts being added to a git repository (without history)
<courmisch>
bonus points for including custom build programs as x86 ELFs
quietvoid has quit [Remote host closed the connection]
Marth64 has quit [Ping timeout: 256 seconds]
quietvoid has joined #ffmpeg-devel
Marth64 has joined #ffmpeg-devel
Marth64 has quit [Ping timeout: 246 seconds]
Marth64 has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
averne has quit [Ping timeout: 272 seconds]
<JEEB>
Lynne: rename and also the header ifndef/define being renamed makes it build it seems
<JEEB>
so the av1 decode header gets included from one of the included system headers
<Lynne>
send a patch
<JEEB>
and thus VULKAN_VIDEO_CODEC_AV1STD_DECODE_H_ and friends get defined at the point where the FFmpeg copy gets included
<JEEB>
I do get re-definition warnings for VK_STD_VULKAN_VIDEO_CODEC_AV1_DECODE_SPEC_VERSION etc but at least it builds, I guess?
kurosu has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
<Lynne>
I have 2 arrays, const uint64_t[1][256] and const uint64_t[4][1024]
<Lynne>
is there no possible way to pass them to the same function's argument without the compiler complaining or casing?
<courmisch>
nope
Marth64 has joined #ffmpeg-devel
<Lynne>
so we can have atomics, threading, alignment, genrics, bitwise ops, utf-8 strings, but we can't pass 2 double arrays with defined different sizes to the same function?
<courmisch>
that's a feature
<courmisch>
the compiler knows that the two arrays can't alias
<Lynne>
meh, double arrays are for losers and mathematicians
<JEEB>
in theory fixing the system header inclusion would be preferable
<JEEB>
right, vulkan_core.h includes the system av1 header
<JEEB>
vOv
<JEEB>
yea so hwcontext_vulkan.h -> <vulkan/vulkan.h> -> vulkan_core.h -> vk_video/vulkan_video_codec_av1std.h and friends
<JEEB>
so at the point of vulkan_video AV1 inclusions in libavcodec/vulkan_video.h you're already including the system one
<JEEB>
*ones
<JEEB>
sent the patch to the list, now for some sleep dess
Marth64 has quit [Ping timeout: 268 seconds]
<JEEB>
(personally if the external API surface or ABI doesn't break, I would not be too much against just backporting the switch to standard headers - as the Mesa defined stuff would most likely at some point Soon just stop working.)
<Lynne>
sure, me too
<Lynne>
but nvidia have shit drivers and jkqxz want a second implementation
<Lynne>
which exists, it's mesa's av1 for intel!
paulk has quit [Ping timeout: 260 seconds]
paulk has joined #ffmpeg-devel
paulk has quit [Changing host]
paulk has joined #ffmpeg-devel
tufei__ has joined #ffmpeg-devel
tufei_ has quit [Ping timeout: 255 seconds]
tufei__ has quit [Remote host closed the connection]
tufei__ has joined #ffmpeg-devel
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 264 seconds]
rvalue- is now known as rvalue
averne has quit [Ping timeout: 264 seconds]
ccawley2011 has quit [Read error: Connection reset by peer]
averne has joined #ffmpeg-devel
averne has quit [Ping timeout: 260 seconds]
averne has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
<Daemon404>
Lynne, cant have locale independent stdlib funcs either.