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
zsoltiv_ has quit [Read error: Connection reset by peer]
zsoltiv_ has joined #ffmpeg-devel
<CounterPillow>
has anyone in the moronix comments called WHIP "woke" yet
<minimal>
not yet, they're all asleep currently lol
ultrazane123 has joined #ffmpeg-devel
ultrazane123 has quit [Client Quit]
<fflogger>
[editedticket] CosmicSkye: Ticket #9996 ([ffmpeg] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:17
minimal has quit [Quit: Leaving]
Anthony_ZO has joined #ffmpeg-devel
_whitelogger has joined #ffmpeg-devel
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 272 seconds]
jamrial has quit []
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<fflogger>
[newticket] gamer191: Ticket #11627 ([undetermined] Feature request: copy IAMF stream_groups from input file) created https://trac.ffmpeg.org/ticket/11627
mkver has quit [Ping timeout: 248 seconds]
ngaullier has joined #ffmpeg-devel
<fflogger>
[editedticket] BlueWindy: Ticket #9996 ([ffmpeg] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:18
<ramiro>
compn: with centos 7 + devtoolset you can ship static binaries that will work in a huge amount of distros.
<TheVibeCoder>
hmm, i use mostly full screen windows. but anyway this niri with scrollable wm is real fun
<pross>
TheVibeCoder: evening!
<TheVibeCoder>
pross: your welcome!
<pross>
the most wanted codec is ac4
<TheVibeCoder>
not really, i see not torrents encoded with it
<TheVibeCoder>
also most AC-4 streams are now 100% DRM
<TheVibeCoder>
and broadcasts
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 248 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
<beastd>
haasn: what are the problems with sway? (never used it, just curious)
<haasn>
beastd: it calls itself a tiling window manager but doesn't know how to tile
<haasn>
and every time people have sent PRs to fix it, the response by the devs was dismissal along the lines of "EDONTCARE"
<beastd>
i must be missing sth, but some sort of tiling should be supported, right? or you are using a different definition for tiling?
j45_ has joined #ffmpeg-devel
<TheVibeCoder>
tile definition is in tile video filter
<haasn>
beastd: well, binary space partitioning would be a nice start
j45 has quit [Ping timeout: 248 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
<haasn>
but infinite scrolling is definitely superior :)
<haasn>
also correction: they used to have bsp but it was removed to "encourage people to work on alternatives"
<beastd>
haasn: yeah :) scrolling looks interesting. not sure i should like it, but i my guts tell me i would.
<haasn>
niri also does other things right, like arranging workspaces vertically
<TheVibeCoder>
does it have fancy animations all way around?
<haasn>
so navigation between workspaces is just scrolling up/down, complementing the left/right motion of navigation within a workspace
<TheVibeCoder>
but i want instant jump
<haasn>
TheVibeCoder: sure, though I recommend to set them to 400% speed to make it usable
<TheVibeCoder>
scrolling is just for shows
<haasn>
at 400% the are brief enough to not be distracting but still provide important contextual clues to your brain versus windows just snapping to their new position
<haasn>
important when you can't see all windows onscreen at once imo
<TheVibeCoder>
pross: i merged your branch, just for the info, if you have not logged in github... - do you have more fixes?
TheVibeCoder has quit [Quit: Client closed]
TheVibeCoder has joined #ffmpeg-devel
linkmauve has joined #ffmpeg-devel
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
ngaullie has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
Anthony_ZO has quit [Ping timeout: 260 seconds]
ngaullier has quit [Quit: Leaving]
Guest89 has joined #ffmpeg-devel
Guest89 has quit [Client Quit]
microlappy has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
<Traneptora>
ramiro: I'm sending a patch to v4l-utils first for now
<Traneptora>
since it's inherently an upstream problem
<Traneptora>
if they don't merge it, then I'll change it in our code
<BBB>
ePirat you said the magic words, woohoo
<BBB>
go gitlab
* BBB
runs
IndecisiveTurtle has joined #ffmpeg-devel
<ePirat>
at this point I feel like a broken record when we talk about this every year over and over again :D
<BBB>
me too, man, me too
<BBB>
it's true though, the mailinglist is so utterly broken and dumb but some people just can't wrap their head around that. sad
<BBB>
I'll just let it simmer and boil over for another few years until it explodes in a fantastic fireworks and out will come ... gitlab, maybe?
<ePirat>
I keep missing patches because sometimes they cant be delivered to gmail addresses...
<BBB>
they go to my spam, but yes
<ePirat>
my main issue is keeping track of new revisions and which comments had been addressed or not...
<BBB>
yeah it's very tricky indeed
<ePirat>
its not perfect in gitlab / forgejo either but still better than having to hunt down old emails
<BBB>
some people still ride horse-and-carriage in NYC, so we're not alone in being backwards
<BBB>
admittedly they're tourists and just-married couples, but I guess we're in good company
<jamrial>
what's stopping deploying the existing forgejo instance right now?
<kierank>
Can we mark WebRTC as experimental
<BBB>
jamrial: from my perspective I still don't understand why we're going with the minix of project management solutions when the correct and widely used solution is right there... but maybe that's not worth the fight anymore
<jamrial>
BBB: afaik, nothing has been decided. it could be foregejo or gitlab
<BBB>
if the goal is to attract new people, we should use something people use. I have never heard of forgejo outside this community
<jamrial>
the current forgejo instance is a test one. eventually, a gitlab one would be set in theory
<jamrial>
then one has to be chosen with a vote
<jamrial>
BBB: it's a gitea fork
<JEEB>
BBB: I think the biggest thing that went for forgejo so far was fedora. they have not started their migration, so we'll see how it goes
<kierank>
BBB: +1
<jamrial>
we need to deploy forgejo now, use it for a while, then do the same with gitlab
<kierank>
Let's be honest gitlab has to be setup by one of Michael's buddies
<jamrial>
then vote and be done with it
<kierank>
And none of them are gonna do that
<jamrial>
why?
<jamrial>
to the first statement
<kierank>
Because otherwise we could just use videolan and be done
<JEEB>
I think BtbN was the one to set up forgejo, and the reason why was that no-one complained about forgejo while there was one vocal voice against gitlab
<TheVibeCoder>
BBB: -1, forgejo comment
<kierank>
I complained extensively about forgejo
<jamrial>
i repeat, the idea is to try BOTH
<haasn>
I'm sure the set of (people who would prefer a self hosted server) x (people who are not Michael's buddy) is nonzero
<jamrial>
one was set up, now it needs to be deployed
<wbs>
jamrial: so you mean the source of truth repo is still where it is, and we do throwaway reviews on both systems for a while to test things out?
<BBB>
meanwhile, I'm gonna write code again
<BBB>
I think I'm where wbs was yesterday, I have much more fun coding than with this stuff, sadly
<kierank>
Gitlab is proven already, it has advanced CI
<jamrial>
BBB: that's everyone
<haasn>
same, I am also just gonna continue coding while I wait for upper management to make a decision
<kierank>
Or we can use forjego with an anime character
<kierank>
People on HN complain they can't get past it
<haasn>
I get the same anime character on gitlab fwiw
<kierank>
On videolan gitlab?
<haasn>
freedesktop
<ePirat>
kierank, the anime character is not a forgejo thing
<kierank>
I understand but it's comparing the videolan vs forgejo setups
<kierank>
Emails don't arrive with forgejo for several people
<jamrial>
wbs: we enable PRs on the forgejo/gitlab instances, and during testing they are kept in sync with the existing git repo
<devinheitmueller>
Also helpful that lots of companies use gitlab, so it lowers the barrier to entry for new developers if they already happen to work somewhere that has embraced gitlab. Good luck saying the same for forgeio.
<jamrial>
kierank: is that a forgejo problem, or just the server's ip not being whitelisted?
<kierank>
Who knows, point is videolan has this all sorted
<devinheitmueller>
Agreed it’s kind of a pity ffmpeg won’t simply use the videolan gitlab instance. Then they get all the maintenance for free since it’s already being done anyway.
<ePirat>
devinheitmueller, tbh though if someone is familiar with gitlab they will probably have no issue figuring out forgejo either, its very similar
<jamrial>
forgejo is more like github tbh
<ePirat>
my main issue with GitLab is their enterprise edition sometimes gating even trivial features
<ePirat>
and some frequently requested seemingly basic stuff not happening, like the ability to have several artifacts for a CI job...
<kierank>
Ci on forjego is really basic
<devinheitmueller>
Any self-hosted solution means somebody will have to own maintenance. Using videolan’s means no additional manpower has to be put toward maintaining the server, which is nice in a project that already has limited resources to manage infrastructure.
<ePirat>
Anyway I feel like we get ahead of ourselves, the first step would be to have a basic vote to decide to go with a git forge, after that we can spend more resources on figuring out which and where...
<ePirat>
if we try to solve every detail before deciding on the most fundamental question, we get nowhere, as evident by the past years discussions about this over and over again
<BtbN>
I'm also actually against Gitlab and in favour of Forgejo. Gitlab has proven to be a constant source for security problems for one. Then it's extremely resource intensive and slow to use.
<BtbN>
And I also feel like it has a very unintuitive interface.
<BtbN>
So if I had to pick between the two, I'd pick Forgejo
<jamrial>
i have noticed videolan's gitlab instance being slow, yes
TheVibeCoder has quit [Quit: Client closed]
<jamrial>
but regarding security issues, it may have to do with the fact gitlab is more widely used and such more likely to be a target :p
<BtbN>
in the end I don't care enough to block either though. I just want to finally get somewhere
<ePirat>
anyway how would we go about organizing such a vote? IIRC Anton used to do these...
<BtbN>
Though I'd estimate hosting cost for Forgejo to be ~20€ a month, while Gitlab would be closer to 50-100€ to reach the same performance
<jamrial>
ePirat: lets deploy forgejo and test people doing PRs with it first
<BtbN>
I think Gitlab being as slow as it is mostly comes down to a) Ruby and b) Bloat
<BtbN>
The thing is, there is not really a good way to "test-deploy" it without it being the source-of-truth repo
<BtbN>
that's just merge-conflict hell waiting to happen
<ePirat>
exactly
<ePirat>
anyway my point was not to have a vote on _what_ to use yet but just a vote that we _do_ want to use _some_ git forge, be it gitlab or forgejo...
<ePirat>
just to get out of the way that "no one wants this"
<jamrial>
BtbN: we disable pushing in git.ffmpeg.org, temporarily setting it as a mirror for forgejo
<BtbN>
the current source of truth is still videolan
<jamrial>
then that one
TheVibeCoder has joined #ffmpeg-devel
<BtbN>
The problem with setting it as a mirror is that gitolite can only mirror with other gitolite
<jamrial>
great..
<BtbN>
I looked into that extensively, and the conclusion was that it's not possible
<jamrial>
so, manual sync only?
<ePirat>
BtbN, you can have forgejo push to gitolite though, no?
<BtbN>
Though we could revoke push access for all but one key, and then run a cron job somewhere that keeps it up to date
<BtbN>
effectively manually mirroring it, like Github
<jamrial>
that sounds fine
<BtbN>
My initial idea was to forward all pushes to it to Forgejo
<BtbN>
but that fails on the simple fact that for that, the paths on both domains would have to match
<BtbN>
and they don't
<BtbN>
maybe some wrapper shell-script could still make that possible, by rewriting the path in every git command, but eh
<ePirat>
IMHO a cron that syncs it every few minutes would be fine
<ePirat>
maybe have a git hook that writes some magic file so the cron can just do nothing when nothing changed
<Traneptora>
well if it's rsync then if nothing changes we got minimal overhead anyway
<BtbN>
The advantage of the push-forwarding is that people who don't care won't notice a difference, and don't need an account
<BtbN>
rsync?
<BtbN>
It's just git
<Traneptora>
if we're syncing with git push then it it's not a lot of overhead to push an already up to date thing
<Traneptora>
"cron can do nothing when nothing changed" doesn't seem that important
<BtbN>
I don't understand what exactly you are talking about
<Traneptora>
I'm responding to >maybe have a git hook that writes some magic file so the cron can just do nothing when nothing changed
<BtbN>
I'm talking about forwarding pushes at the SSH level, so people can keep working as they are right now, despite a new source of truth
<Traneptora>
I'm saying that I don't think this is necessary
<Traneptora>
the thing I quoted
<BtbN>
I feel like it actually is, since otherwise we will have a heapload of people complaining about having to make an account
<BtbN>
Making friction as low as possible has merit
<Traneptora>
the thing I quoted, I meant "so cron can do nothing when nothing changed"
<Traneptora>
I don't think that specific thing is necessary
<BtbN>
that's a non-issue anyway, a pull&push when nothing changes is almost free
<Traneptora>
that's exactly what I said
<BtbN>
It's what the github mirror is doing every few minutes since ages
microlappy has quit [Ping timeout: 248 seconds]
<Traneptora>
we're on the same page then, I'm just commenting on that one suggestion
TheVibeCoder has quit [Quit: Client closed]
TheVibeCoder has joined #ffmpeg-devel
<kierank>
14:41:12 <•BtbN> Though I'd estimate hosting cost for Forgejo to be ~20€ a month, while Gitlab would be closer to 50-100€ to reach the same performance
<kierank>
we have hundreds of thousands in the bank, let's pay to have proper infrastructure
<kierank>
not things held together on a 20eur hetzner that needs anime characters to be stable
<BtbN>
get a grip man
<kierank>
I mean seriously, the infrastructure (by design of the "leadership") is held together with string
<kierank>
and you seem to celebrate this
<fflogger>
[editedticket] Balling: Ticket #9996 ([ffmpeg] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:20
<BtbN>
Whatever your issue with Hetzner is now. It's a perfectly normal hoster.
<kierank>
this isn't a high school project
<BtbN>
So you want us to build or own datacenter or what are you saying?
<TheVibeCoder>
this is a college project
<BtbN>
There is absolutely nothing wrong with Hetzner as a hoster.
<kierank>
15:32:08 <•BtbN> So you want us to build or own datacenter or what are you saying?
<kierank>
yes that's exactly what I'm saying...
<BtbN>
We have a few million too few in the bank for that lol
<kierank>
that's sarcasm of course
<kierank>
but videolan is a comparable size project, proper servers, proper datacenter, sysadmin, emails work, no anime characters
<BtbN>
And dead slow Gitlab...
<BtbN>
And Videolan is going to be using some hoster as well...
<kierank>
looool
<kierank>
oh no the pages load 500ms later, but we have actual CI
<kierank>
and not the third fork of a git repo
<kierank>
(I know, we are using gitea at work)
<BtbN>
I'm this close to just ignore anything you say from now, cause you behave like some bratty child
<kierank>
seriously
<kierank>
<•BtbN> Though I'd estimate hosting cost for Forgejo to be ~20€ a month, while Gitlab would be closer to 50-100€ to reach the same performance
<kierank>
this is peak child
<kierank>
ffmpeg is a serious piece of global infrastructure
<kierank>
not a bedroom project
<BtbN>
well, welcome to my ignore list then. I don't care anymore about the permanent nonsense attacks.
<kierank>
anime characters on serious infrastructure
<kierank>
truly people wanting to make this project a joke
<kierank>
and reveling in the strings holding it together
<jamrial>
kierank: suggest to pay to remove the anime character
<jamrial>
like other users of anubis did
<kierank>
or have hosting done in a place that doesn't need it
<kierank>
omg
<jamrial>
really now?
<kierank>
you're treating the symptom not the ailment here
<Gramner>
trac.ffmpeg.org is the #1 result when searching for ffmpeg trac here
<michaelni>
jamrial, emma is in MAINTAINERS and has git push access. Is there any problem ?
<TheVibeCoder>
google provides personalized results since "ever"
<kierank>
TheVibeCoder: use baidu like your friend kostya
<kierank>
TheVibeCoder: when will librempeg acquire ffmpeg
<TheVibeCoder>
when whip protocol is feature-complete
<another|>
when the chains are added /s
kasper93 has joined #ffmpeg-devel
<compn>
ramiro, centos7 EOL june 2024.
<kierank>
TheVibeCoder: imo NACK is pretty mandatory
<kierank>
otherwise it's useless outside a lan
<kierank>
and isn't "web"rtc
<compn>
ramiro, but yeah its unimportant for making builds
<fflogger>
[editedticket] Balling: Ticket #10374 ([avcodec] Problems to decode certain h264 keyframes with D3D11VA and DXVA2 hw acceleration) updated https://trac.ffmpeg.org/ticket/10374#comment:11
ngaullie has quit [Remote host closed the connection]
mkver has quit [Ping timeout: 244 seconds]
softworkz has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
<softworkz>
Lynne: "I'm not sure I want to keep contributing with softworks having push rights" << would it make you feel better when I would ask you personally each time before I push anything? (it's a serious question)
<Lynne>
I'd rather see you simply stop making questionable patches
<Lynne>
all other developers, even the ones with years less experience, know what ideas are worth sending to the mailing list
<softworkz>
Sure, but what do you think about that idea?
<Lynne>
if you call system() from ffmpeg.c, it's probably a bad idea
<Lynne>
if you disagree with every single developer, its probably a bad idea
<softworkz>
It was meant as a serious proposal.
<softworkz>
To consult you before I push anything
<Lynne>
the fact that you're even asking about this is a bad idea
<softworkz>
Ok thanks
softworkz has quit [Quit: Leaving]
<Lynne>
jeez, none of us are a review-as-a-service over HTTP APIs
softworkz has joined #ffmpeg-devel
<softworkz>
It wasn't about reviewing
softworkz has quit [Client Quit]
TheVibeCoder has quit [Quit: Client closed]
TheVibeCoder has joined #ffmpeg-devel
<Lynne>
we're not a sign-off-as-a-service either, if a patch were to slip through the one signing off would probably be the one who has to fix it
Flat has quit [Quit: Rip internet]
Flat has joined #ffmpeg-devel
softworkz has joined #ffmpeg-devel
lemourin has quit [Ping timeout: 252 seconds]
softworkz_ has joined #ffmpeg-devel
softworkz is now known as Guest4361
softworkz_ is now known as softworkz
Guest4361 has quit [Read error: Connection reset by peer]
<softworkz>
It isn't meant to employ you for anything. It was merely about letting you have a final view after everything else
<softworkz>
When my commit rights are worrying so much that you think about quitting, then I'm wondering why the ability to oversee my pushes wouldn't be a suitable measure for re-establishing trust?
TheVibeCoder has quit [Quit: Client closed]
<softworkz>
I'm not trying to make any points here. It is about the last word in the previous message. It's just a suggestion because I'm not happy to see such extreme impact I'm appearing to make. Pleople not liking me is one thing. But I don't want to get that far like it seems to be now.
TheVibeCoder has joined #ffmpeg-devel
<softworkz>
Just think about it please. Thanks
softworkz has quit [Quit: Leaving]
<TheVibeCoder>
unstable weeb
<jkqxz>
Because of the implied obligation to do something, which many open-source contributors absolutely don't want. You would (quite reasonably) complain if said overseer is not very interested and took a very long time to respond.
lemourin has joined #ffmpeg-devel
<kierank>
TheVibeCoder: anton leaving opened floodgates of craziness
<kierank>
forejgo, system() pushed without review, webrtc pushed without review
<TheVibeCoder>
wild west
<kierank>
TheVibeCoder: librempeg is bastion of stability
<jamrial>
webrtc was pushed without review?
<TheVibeCoder>
it was reviewed internally
<jamrial>
also, anton disliked gitlab, lol
<kierank>
04:41:36 <mkver> Did someone even review this? It adds both a new protocol and a muxer in the same commit.
<kierank>
jamrial: yet he's using librempeg github
<kierank>
tbh the Linux distro approach is maybe a workable future for ffmpeg
<jamrial>
yes, i said gitlab. he disliked how it worked when he looked at videolan's a few years ago
<kierank>
ok fair point
<jamrial>
like how an MR originating from a private repo would report errors when fetching CI artifacts instead of just saying "Nope, can't show this"
softworkz has joined #ffmpeg-devel
IndecisiveTurtle has joined #ffmpeg-devel
<softworkz>
jkqxz: Lynne: Sorry for being unclear: It isn't meant to acknowledge, it is about giving a chance to object and without any responsibility. Advice is always welcome, like "xy should take a look at it" or "wait for another week" or so. But this is not about giving okay. Only about the chance to nack. It is very clear that it would be asked way too much otherwise.
softworkz has quit [Quit: Leaving]
<TheVibeCoder>
get stable web for once
<TheVibeCoder>
do not leave ever
lemourin0 has joined #ffmpeg-devel
lemourin is now known as Guest4525
Guest4525 has quit [Killed (zinc.libera.chat (Nickname regained by services))]
lemourin0 is now known as lemourin
<another|>
re: whip: it also adds dtls in the same commit if I read that correctly
<JEEB>
yes, and openssl only
<JEEB>
and I think dtls is something something UDP only?
<JEEB>
I wonder if that thing works with mediamtx
<JEEB>
or if it's only tested with some very specific media server
<JEEB>
or service
<another|>
I think DTLS is TLS for UDP
<another|>
just a casual 2.9k LOC
<kierank>
TheVibeCoder: you used to have unstable net
<TheVibeCoder>
kierank: your gorgeous donations fixed it all!
<kierank>
elon gave you starlink?
<TheVibeCoder>
watch your own business!
<TheVibeCoder>
kierank: be really constructive and complete whip code
<kierank>
not possible without event loop
<kierank>
does ffmpeg support bidrectional udp sockets
lemourin2 has joined #ffmpeg-devel
lemourin is now known as Guest3695
Guest3695 has quit [Killed (lead.libera.chat (Nickname regained by services))]
HarshK23 has quit [Quit: Connection closed for inactivity]
TheVibeCoder has joined #ffmpeg-devel
softworkz has joined #ffmpeg-devel
TheVibeCoder has quit [Quit: Client closed]
<softworkz>
TheVibeCoder: "get stable web for once" << in case you meant my leave, I'm sorry, it's no disrespect or so. I just can't have an IRC client running when doing focused work. It's stupid, but I always need to look at it when it's on..
<fflogger>
[editedticket] jamrial: Ticket #9996 ([ffmpeg] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:21
softworkz has quit [Quit: Leaving]
microchip_ has quit [Ping timeout: 248 seconds]
microchip_ has joined #ffmpeg-devel
<fflogger>
[newticket] berarma: Ticket #11628 ([avformat] Ogg: Broken file, keyframe not correctly marked) created https://trac.ffmpeg.org/ticket/11628
IndecisiveTurtle has quit [Read error: Connection reset by peer]