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
<BtbN>
There is primarily a company behind it, not so much a community.
<BtbN>
Which is both good and bad
<kierank>
There is freedesktop, videolan etc
<BtbN>
Which all depend on the Company
<kierank>
Much bigger than a few purists with forgejo
<BtbN>
If they suddenly get crazy ideas, both Freedesktop and Videolan will be at their whim
<kepstin>
getting issues fixed in gitlab is a real pain unless you have a corporate support contract, which can push issue priority up :(
<rcombs>
I think if I had to rank the options right now, I'd probably say github > gitlab > forgejo > ML
<rcombs>
but gitlab vs forgejo could plausibly flip in a year
<beastd>
kierank: that means no other forge should ever be used? no one used ffmpeg in the beginning. no one bet on ffmpeg. arpi at some point did. what followed is a huge success for ffmpeg.
<rcombs>
but honestly I don't think we're going to have huge regrets with either
<kierank>
beastd: false equivalency
<kierank>
We are a piece of internet infrastructure, we shouldn't take risks (as we do now) with our own infrastructure
<beastd>
kepstin: personally i'm skeptical to most "actions". IMHO we should use as few as possible and most stuff should just be in the code base and hooked into from simple "run shell" action
<rcombs>
I think forgejo will probably feel less risky in a year than it does today, but today it's legitimately a concern, even if it's not a huge one
<kepstin>
if you specifically want "infrastructure that can be trusted by fortune 500 companies interested in verifying their supply chain", then github is certainly the best option, simply because they see "Microsoft" and go "yep, sounds good" :/
<kierank>
I would have said the same about gitea
<rcombs>
and I'd rather not wait the year
<kierank>
But now I'm stuck
<kierank>
Same with gogs 5 years ago
<kierank>
Both looked fine
<kierank>
They work well enough for the small repos at work
<kierank>
But that's it
<compn>
code.google.com was shutdown. big companies shut down projects all the time
<beastd>
kierank: not sure i see where you are aiming at. wouldn't ffmpeg have had grown so much it could now have freedom?
<kierank>
That's the past
<BtbN>
Neither Github or Gitlab are safe either
<kierank>
We have responsibilities now
<kierank>
As internet infrastructure
<BtbN>
The Gitlab-Company could go rogue due to an acquisition or something, or Microsoft could turn Github into AI-Hell
<kierank>
Otherwise chuck system() in and webrtc
<beastd>
and there was a lot at code.google.com also at freahmeat and also at sourceforge (which surely is one of the earliest widely used forges)
<rcombs>
on the other hand, I'd strongly prefer something self-hosted over videolan's *unless* they get the signup situation resolved
<rcombs>
(…but if they do resolve the signup situation, then I'd prefer to leave it to them)
<kierank>
freshmeat was an aggregator
<kierank>
Sourceforge arguably for big binary projects is a good choice
<kierank>
The mirrors are still the best
<rcombs>
beastd: didn't google code die because github came around and was very obviously better so everybody moved
<beastd>
kierank: yes. freshmeat was sth differen. my fault
<BtbN>
Google Code died cause Google decided so
<kierank>
If I needed to ship 5GB binary in my project sourceforge is still the best for thst
<BtbN>
They just made it "Google Projects Only" one day
<rcombs>
well yeah, but there were reasons that they decided it wasn't worth investing in anymore
<beastd>
rcombs: maybe. maybe not. does it meant no one should have tried to use it?
<rcombs>
and a lot of that came down to github eating its lunch
<kierank>
It's even worse, you could have Gogs vs Gitea vs Forgejo vs Xxx?
<kierank>
Onedev as well apparently
<kierank>
Never heard of that one
<beastd>
also i personally was much more positive about google 20 years ago. no one can see what the future holds. and of cource meta, google and microsoft still do nice things
<rcombs>
compn: so they even specifically cite github
<compn>
rcombs, yep. they also cite how they were spending amounts on spam prevention.
<beastd>
kierank: i have used gogs very long now (you were the one who introduced me to it).
<kierank>
Yes Gogs worked fine, problem is it has forked twice now and in a non backwards compatible way
<kierank>
So you have to make consequential decisions
<beastd>
kierank: i was very sceptical about gogs in the beginning. but after using it for some months i must say it really did many things right. it had this initial momentum that worked and had a good balance. i read a lot of the gogs source code by now.
<beastd>
kierank: the fatal flaw with gogs was, it was and is extremely undermanned
<beastd>
gitea did a lot better on this and problems that are very important for many use cases have not been adressed despite being obvious and some reported for 1 or 10 years ago
<kierank>
I don't think it has the level of sophistication a large project like ffmpeg needs
<kierank>
We have some strange bugs in Gitea as well
<compn>
rtmpdump was dmca'd off sourceforge in 2009 (and hosted with ffmpeg, since)
<beastd>
gitea mostly solved it, but then they had a hiccup about goint into "company owned" situation which cause the fork that is name forgejo (and yes didn't like that name much)
<compn>
not sure how many people still use librtmp though. i guess people still use it
<beastd>
ffmpeg has some strange bugs and linux kernel has strange bugs, glibc had strange bugs, bugzilla had strange bugs... question is what and how was it resolved and in how much does it matter to the decision at hand.
<beastd>
all real world software has "technical debt" (don't like the term but it's the most popular)
<kierank>
It matters because PRs were broken and I had to do some weird undocumented hack to make it work
<beastd>
kierank: understood. didn't trip upon that. gitlab also had critical issues. didn't trap on them neither.
<beastd>
but sure if PRs in forgejo were broken "regularly" that would be an argument against it
<BtbN>
The Gogs/Gitea/Forgejo Codebase is pretty approachable, so fixing bugs is very realistic
<BtbN>
good luck trying that with Gitlab, getting anything through there is a nightmare from what I heard
<beastd>
i can also say that for gogs/gitea/forgejo (so no idea how easy it is to get in, especially in gogs). no idea about gitlab. also i'm probably not familar with the implementation languages.
<kierank>
01:16:57 <• beastd> but sure if PRs in forgejo were broken "regularly" that would be an argument against it
<kierank>
They were broken because I had upgraded from gogs
<kierank>
I think clean install would have worked
<beastd>
kierank: ok
<BtbN>
The mail bug our instance had run into was also fascinating. I must have managed to restart the instance exactly in the middle of it bumping the mail queue
<BtbN>
leaving the internal queue in an invalid state, where it reported empty at all times
<beastd>
switching between gogs far apart versions and between gogs and gitea and between gogs and forgejo is painful and fragile
<BtbN>
Fixing it was as simple as hitting the "Flush Queue" button
<beastd>
at least forgejo has something called an f3 implementation in go that works with some gogs, gitea and some more systems.
<beastd>
for completely round-tripping that is
<kierank>
01:31:45 <• beastd> switching between gogs far apart versions and between gogs and gitea and between gogs and forgejo is painful and fragile
<beastd>
but from my experience it's not fully (straight-forward) usable. would be good if that could be improved. best with more funds probably.
<kierank>
Yes and that's likely to continue looking at the past
<kierank>
I don't know wjether to continue with Gitea or move to forejgo at work
<beastd>
most modern projects with persistency of this kind do not support large hops when upgrading versions
<kierank>
As well as having to explain to people what a forjego is.
<beastd>
kierank: not sure for your use case. you should think about it now tho because the soft fork phase of forgejo has ended
<kierank>
If it goes wrong I think I can't go back
<kierank>
Without backup etc
<BtbN>
it's still possible to migrate up to date gitea to forgejo
<compn>
backup
<kierank>
And I run this on a Synology NAS which makes it more complex
<beastd>
for new projects and installations i prefer forgejo because of the organization (not because of the name, tho that as proven to be not as big a as a problem as i thought it would for communications, after metioning it a few (pronounced partially wrong) time it just stuck :)
<BtbN>
People have made migration scripts/guides
<BtbN>
it just doesn't work automatically anymore, but it's doable
<kierank>
Unlike ffmpeg, we don't just do random jank scripts at work
<kierank>
Like it costs a lot of money if git push is down
<beastd>
kierank: i think for switching back and forth between forgejo and gitea it should work with f3. i might be finicky tho.
<kierank>
That's my issue
<BtbN>
It's probably also possible to migrate from Forgejo to Gitea if one is motivated enough
<BtbN>
the DB layout is still super similar
<kierank>
You're proving my point here
<kierank>
That's anything is possible but nothing is certain
<BtbN>
Well, you'd just test it on a test instance, document the procedure, make backups, and then execute it for the production one
<BtbN>
I don't see the issue, we do stuff like that regularily
<beastd>
kierank: Maybe try to run it parallel with some kind of sync for some time. that should be safest bet. of course depends on your company and working habits if that could fly?
<kierank>
I can't, it's on a Synology NAs
<BtbN>
Why tho
<kierank>
Id have to buy a new physical NAS
<kierank>
Everything is tied into the way they do it
<beastd>
kierank: can't it run those in different container apps on different ports?
<kierank>
It uses a global mysql
<BtbN>
tbf, I don't think Gitea is about to go anywhere, so nothing wrong with keeping to use it
<beastd>
oh :-/
<kierank>
01:39:29 <• BtbN> I don't see the issue, we do stuff like that regularily
<kierank>
Ffmpeg sysadmin is not something to emulate
<kierank>
I know who owns the domains at work
<BtbN>
I'm talking about work...
<kierank>
Nobody will be honest in this project about who owns avcodec.org
<BtbN>
You got the honest answer already.
<kierank>
Although I can guess
<kierank>
Who owns avcodec.org and why is it not documented?
<BtbN>
Michael thinks it's more secure if it's unknown, anti social engineering
<BtbN>
I don't agree with that, but that's how it is
<kierank>
Lol
<kierank>
Personality cult open source
<compn>
with the amount of probing kierank has done to our hosts and domains...
<BtbN>
I can't exactly force it out of him, not sure what about that is "personality cult"
<kierank>
The fact you just kowtow
<kierank>
And not see it as a problem
<BtbN>
I literally don't know either
<kierank>
One of the most important open source projects has an unknown domain owner
<compn>
still dont know why its so important. and you refuse to answer why
<compn>
maybe i got ignored? :D
<BtbN>
Anything that doesn't fit the narrative is ignored I feel
<BtbN>
By multiple people involved
<compn>
i thought our github thing could push patches to our ml?
<beastd>
sorry it's really late here... thank you all for the conversation. see you probably tomorrow at some point. have great time of day wherever you are!
<compn>
pr i mean
<BtbN>
no, that's Softworks personal thingy
<compn>
ah
<compn>
i dont know how to help softworkz. seems like a lot of personalities are clashing. maybe this project isnt good for him
<BtbN>
My impression is that he's just not good with any kind of criticism, and takes it all as personal attacks, and as a result goes on exhausting rants that people then just tune out of
<compn>
i wonder if some of it is due to the developers having so much experience , that the skills/patience for dealing with newbies and newbie type contributions is gone.
<compn>
or that what you said
<compn>
certainly the patience for long mails is long gone :)
<BtbN>
Like, multiple people have given substantial and in depth technical cricisim of his Subtitle-Filtering Approach
<BtbN>
but it's completely shot down by him, so it's just him ranting against people going "Hard NAK"
<michaelni>
about the subtitles, I think nicolas should provide simple testcases
<BtbN>
I mostly think the technical concerns posted multiple times should be taken seriously
<compn>
providing test samples is helpful to newbies. but antagonizing developers reviewing patches is not good
<compn>
also i'm annoyed by people who immediately move to legal terms like defamation.
<compn>
and certainly its impossible to work with people who take or make legal threats
<BtbN>
Yeah, I was apparently also defaming him cause I didn't find him addressing reviews in a 10-15 layers deep thread I didn't even realize exists.
<michaelni>
what i meant was, nicolas claimed there are specific shortcomings in the patches design which softworks claims these are not there. I just tried to suggest on the ML that nicolas would provide testcases as it would clarify that one aspect. it doesnt solve other peoples concerns
<BtbN>
Multiple people pointed out very valid shortcomings and issues
IndecisiveTurtle has quit [Quit: IndecisiveTurtle]
<BtbN>
but nothing of it is taken into account, instead it's being discussed away with many words
<compn>
er take legal action / make legal threats*
<michaelni>
I have not followed this patchset in enough details to really argue about the specific concerns. I just noticed the case where issues where pointed at and no testcases where provided which felt a bit ... inefficicient
<compn>
the maintainer said he reviewed it and rejected the patch as-is
<compn>
which is the end of story
<michaelni>
compn, where was there a legal threat ?
<michaelni>
yes :( this was a unfortunate series of events. Someone misreads a mail makes a tiny comment on IRC that gets repeated multiple times on the ML, and then its repeated all over the place. This makes sw look very bad and he complains through these mails which noone likes
<michaelni>
Noone had any bad intent. i believe but everyone seems to really be unhappy now
<fflogger>
[editedticket] jamrial: Ticket #9996 ([avformat] Write joc_complexity_index to dec3 (EAC3SpecificBox), Windows and Android need it to play atmos) updated https://trac.ffmpeg.org/ticket/9996#comment:22
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<BBB>
the emails are too long, nobody reads that much stuff
<softworkz>
compn: "take legal action / make legal threats" < it was meant as a reminder about what it actually is but not worth my time pursuing anybody
<softworkz>
BtBN: "My impression is that he's just not good with any kind of criticism" < no problem with criticism, but when someone says something untrue - like you said I hadn't addressed all review comments - that was clearly not true
<softworkz>
compn BtBN: "no, that's Softworks personal thingy" < it's GitGitGadget, that's what Git developers have for posting to their ML from GitHub
<softworkz>
BtBN: "concerns posted multiple times should be taken seriously" < the problem is that so many things have been said, that many do in fact believe that there would be many concerns. When it say it's not true (and one not valid, another one resolved), nobody believes me, and that makes it really hard
<softworkz>
I think the only way to get this drawn out of the quarantine that some have put it into is when another developer gets to FULLY understand the whereabouts and can share her/his view with everybody
<softworkz>
And to everybody: Please criticize me as much as you want - but please please for something that is true.
<softworkz>
At some point I had almost wished I would have done what had been said - then I could have apologized and it would have been done
<softworkz>
Best wishes and good night :-)
softworkz has quit [Quit: Leaving]
<CounterPillow>
is this the killer feature
jamrial has quit []
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 252 seconds]
<compn>
softworkz : no worries . i hope you arent dissuaded too much. sometimes we can be ornery
Anthony_ZO has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
<kepstin>
huh, out of curiousity i tried creating an account on the videolan gitlab. they approved it and i got my account 3 hours later without me having to do anything else. No idea how representative that experience is.
Kei_N has joined #ffmpeg-devel
Kei_N_ has quit [Ping timeout: 248 seconds]
<compn>
>complains about anime anubus
<compn>
>talks to a bunch of anime avatars on twitter
<compn>
ehehe
HarshK23 has joined #ffmpeg-devel
<kode54>
compn: ah, was wondering how softworkz was coming here at random and replying to conversations they weren't present to see, then leaving
Guest1233 has quit [Killed (zinc.libera.chat (Nickname regained by services))]
lemourin2 is now known as lemourin
linjie has joined #ffmpeg-devel
___nick___ has quit [Ping timeout: 265 seconds]
softworkz has joined #ffmpeg-devel
<softworkz>
compn: thanks - all fine :-)
softworkz has quit [Read error: Connection reset by peer]
softworkz has joined #ffmpeg-devel
<softworkz>
kode54: I have a problem when the chat app is running - I always need to look at it and I'm distracting myself...
<TheVibeCoder>
system("kill chat app");
<softworkz>
:-)
<softworkz>
when I've been around the last time (2020-22) there was no logger..
<softworkz>
this year, the first thing I wanted to do is set up a logger
<softworkz>
but luckily there is one again
<llyyr>
killer feature
<softworkz>
LOL
<softworkz>
when you've used it once you don't wanna miss it...
<softworkz>
I thought about externalizing the browser launch into a lib that needs to be installed separately and is loaded dynamically when present:
<softworkz>
libkillerfeature.so
ramiro has quit [Ping timeout: 260 seconds]
softworkz has quit [Quit: Leaving]
ramiro has joined #ffmpeg-devel
<kierank>
haasn: Charles poynton liked your blog post
zsoltiv has joined #ffmpeg-devel
zsoltiv_ has joined #ffmpeg-devel
<haasn>
I should write a blog post about my swscale rewrite
linjie has quit [Quit: Connection closed for inactivity]
<kode54>
softworkz: understandable
<kode54>
alas, I keep chat apps running anyway
<kode54>
could also explain why I get nothing done some days
<BBB>
haasn: you should, that would be super interesting
<kierank>
BBB: +1
psykose has quit [Remote host closed the connection]
psykose has joined #ffmpeg-devel
<BtbN>
Just get a bouncer like most of everyone else.
zsoltiv has quit [Ping timeout: 252 seconds]
<kode54>
I use a cloak to hide the fact that I'm not using a bouncer
zsoltiv_ has quit [Ping timeout: 265 seconds]
zsoltiv has joined #ffmpeg-devel
<beastd>
haasn: +1 would also like to read it :) i unfortunately got lost about the details of your sws rewrite a few months ago, but at least as i understood it so far i'm still hyped :D
<TheVibeCoder>
would usage of floats cause bitperfect issues?
<kepstin>
it can, depending on the operations performed. some float operations are implemented in library code, so the results can change depending on system arch and libraries in use.
<TheVibeCoder>
swscale2 use floats
zsoltiv_ has joined #ffmpeg-devel
<ramiro>
haasn: please do!
Anthony_ZO has quit [Ping timeout: 276 seconds]
jamrial has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
aaabbb has quit [Quit: Free Palestine]
ngaullier has quit [Remote host closed the connection]
ngaullier has joined #ffmpeg-devel
Kei_N has quit [Ping timeout: 245 seconds]
ngaullier has quit [Remote host closed the connection]
ngaullier has joined #ffmpeg-devel
Kei_N has joined #ffmpeg-devel
ngaullier has quit [Remote host closed the connection]
ngaullier has joined #ffmpeg-devel
<kierank>
>gitlab is slow
<kierank>
videolan gitlab loaded in 1.52s
<kierank>
forgejo loaded in 1.61s
<TheVibeCoder>
> 0.5ms
mkver has joined #ffmpeg-devel
<kierank>
for some reason forgejo makes a 2.5s POST request
<compn>
Firefox could be put out of business should a court implement all the Justice Department’s proposals to restrict Google’s search monopoly, an executive for the browser owner Mozilla testified Friday. “It’s very frightening,” Mozilla CFO Eric Muhlheim said.
<compn>
i guess its more mozilla has no faith in itself
<BtbN>
90% or so of their income comes from Google
<BtbN>
I do kinda find it hard to believe that a browser like Firefox couldn't be completely carried by the community though. Surely enough people have an interest in a non-Google browser
<devinheitmueller>
Ultimately the engineers have to feed their families, and the bulk of users of any browser don’t actually pay anything for them.
<Xe>
^
<compn>
google did this monopoly to themselves
<BtbN>
Chrome getting forcibly split out of Google would be intersting though
<JEEB>
thunderbird seems to be now operating mostly by donations towards MZLA Technologies
<compn>
the community can barely keep chromium/firefox forks going. so an entire web browser (some are in development) seems out of reach
<JEEB>
this makes it seem as if the $8.6M of donations are essentially their budget
<JEEB>
if so, it looks like (for now) thunderbird is rolling
<devinheitmueller>
2/3rds of that is personnel. $6M USD really doesn’t go as far as you would like when you’re talking about developer salaries (and loaded costs)
<compn>
email clients dont have to change as much as web though. web changes every day
<compn>
so budget can be funded by donations for sure
<JEEB>
yes, and email clients are generally more "niche" than browsers. but this is in general all up to how mozilla and folks play their cards
<devinheitmueller>
Yeah, it’s not to feel like I would be perfectly happy with a standalone mail client from 2003 (plus security fixes)
* compn
looks at his email client from 2001. err yeah...
<JEEB>
past few years I've seen them not put enough onus on the nice features (tab containers, more lately tab groups and deduplication as well as now with the latest beta tab unloading)
<compn>
JEEB, mozilla has been spending that google money in ways that really didnt do much except burn money. firefox os, buying pocket (which they just closed), starting ai (which they just closed), etc etc
<JEEB>
instead they get into news due to some corporate level generic person not understanding what the general likes and dislikes of their user base (or at least a large enough part of it) are
<BtbN>
Yeah, I also feel like most of that quite a lot of money does just go into nonsense
<JEEB>
compn: re: machine learning the local neural network based translation is <3
<JEEB>
I love the idea of running that sort of stuff locally and not going to translate.xyz.com for it
<JEEB>
it's not as perfect as those "throw lots of compute at it" companies' proprietary things, but the EU-backed (IIRC) projects they based on do actually produce pretty alright results
<JEEB>
but do you see these things being widely advertised? ┐(´д｀)┌
<JEEB>
yea, the translation models etc were funded by the EU
<JEEB>
also a fun bit of that was that I found out accidentally with the Finnish public broadcaster's site that it actually can catch DOM changes such as subtitles, and you suddenly have realtime subtitle translation for videos
<iive>
mozilla has been hostile/indifferent toward users for very long time. The prime example was the elimination of the XUL extensions that allowed to change browser behavior.
<BtbN>
Can't deny that it did make the browser drastically more performant though
<iive>
BtbN, wrong. They switched to another rendering engine at the same time.
<BtbN>
Yes, which wasn't possible before, cause of needing XUL
<iive>
wrong again.
<BtbN>
It's what Mozilla said
<BtbN>
since the old engine rendered the whole browser
<iive>
the xul extension were operational for a while, but only in developers builds
<iive>
yes. they said that. they lied.
witchymary has quit [Remote host closed the connection]
<iive>
aka, hostility toward users.
<BtbN>
Are you sure those builds didn't simply drag both engines along?
<iive>
mozilla still uses XUL for its internal plugins
<iive>
or at least did so years after the switch.
<BtbN>
I just have trouble believing they'd do anything out of malice towards their users. Makes no sense, given those users are their capital.
witchymary has joined #ffmpeg-devel
<kepstin>
they still render the entire browser with their engine. keep in mind that 'XUL' is just a markup language for ui elements. but they've been moving towards using HTML+JS for more elements of the browser UI since people are more familiar with that and it's quicker to iterate on than the C++ code that handled XUL behaviours.
<iive>
well, since XUL accesses internal directly, it imposes maintainers cost when changing internals.
<iive>
so they could fire more developers and move the money to the management.
<iive>
XUL also was not quite friendly to multi-threading. still some plugins were transitioning to the api they proposed.
<kepstin>
not multi-threading, but rather the multi-process model where web content is in an isolated sandboxed process separate from the UI
<iive>
i think that came much later.
<kepstin>
it was a major reason that they wanted to move away from xul extensions tho, to set it up so they could make that change
<iive>
iirc the XUL MT was message based, so it should've been no issue to use it in multi-process instead of multi-thread.
meego has joined #ffmpeg-devel
<pross>
compn: you are correct, g728dec should also depend on celp_filters
<compn>
pross, thanks for yet another decoder :)
<pross>
ei, the celp_math test is always enabled. it fails when we don't enable codecs that drag in celp_math (g729 etc)
microlappy has quit [Quit: Konversation terminated!]
microlappy has joined #ffmpeg-devel
johnjaye has joined #ffmpeg-devel
<compn>
ok. it took me way longer to find git blame until i remembered github had an online git blame :D
<compn>
had to go looking at the original commit for g729
<compn>
and then without looking at the makefile trying to reverse engineer why each codec needed each part. then i just gave up
<compn>
i figured you would immediately know if it was needed or not :D
<compn>
work smarter, not harder.
<pross>
ok. i will add a 'celp_math' CONFIG_EXTRA item to configure, makefile, etc
<compn>
wonder how many people build ffmpeg with only certain codecs
<compn>
i guess its very useful for shrinking binary, easier to load
ccawley2011 has joined #ffmpeg-devel
<compn>
kepstin, the browser using itself to render is one of my peeves for sure.
mkver has joined #ffmpeg-devel
Everything has joined #ffmpeg-devel
<pross>
--disable-everything is also useful for shrinking build times
ccawley2011 has quit [Read error: Connection reset by peer]
ccawley2011 has joined #ffmpeg-devel
bsFFFFFF has quit [Remote host closed the connection]
ccawley2011 has quit [Read error: Connection reset by peer]
<meego>
FWIW we do. We build small static binaries which we run on AWS Lambda. AWS Lambda requires total app bundle size to be <50MB zipped.
<CounterPillow>
sounds impossible for today's webbloat soydevs
<meego>
ahah yes, most node.js devs i know wouldn't know how to tree-shake their node_modules folder
<meego>
but to their credit, the whole ESM/CJS debacle makes this kind of thing way more complicated than it should
Everything has quit [Quit: leaving]
<kierank>
When I reply to the ML I get a forwarded copy back to myself
<compn>
you set yourself to get a copy in mailman prefs ?