DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2021-02-18 00:00:00Z and 2021-02-19 00:00:00Z

[ddnet-maps:master] 1 new commit

6f3051f A airtemple - ddnet-maps

@Learath2 I fixed all places I could where visual studio showed me a high CPU use, even the 128p translation does not loop through every client anymore on CServer::Translate(). That helped a lot already, I can join 128 clients without extreme laggs, but on the production server there are sometimes random laggs, even with fewer people (60-70). The laggs are probably caused by DoSnapshot, because it's usage goes up to 70% on my local server. Thats why I asked whether I can thread it or not, also Mr. Anderson told me I should use threads. So i am asking again: Would it be possible to do so?

10:28

wow, it seems like the 0.7 snapshots are way less efficient...

10:28

i am looking into that now

10:33

maybe that was also false, i dont know

You can't thread snapshots, the main loop cant just go on without snapping. You'd have to lock the main thread

11:36

You need to figure out what part of snapping is slow and fix that

11:41

You could maybe have some benefit from snapping clients in parallel but it's way more work than you'd think to get it all working correctly

11:42

All that work just to check whether it helps or not

11:43

I bet there is more low hanging fruit before that

GetItem() is slow

14:10

But it cant be optimized ig

Image attachment

15:00

https://github.com/crev-dev/cargo-crev

crev-dev/cargo-crev

A cryptographically verifiable code review system for the cargo (Rust) package manager. - crev-dev/cargo-crev

15:00

lol

15:00

xd

actually a cool project

15:14

reminds me of openpgp

15:14

but for projects

Systemd update broke my laptop, sendhelp

18:44

fucking systemd-coredumpd eating 99% my cpu. Such a great service very needed

greenthing

Tbf my own stupidity. I just was too lazy to use gentoo on a laptop with weaker cpu

18:44

Never again I will be lazy about installing a distro without this mess

u mean u were lazy and didnt use gentoo

18:45

or u used it?

Ryozuki

u mean u were lazy and didnt use gentoo

I didn't on the laptop bc it was a little weak

why u install a systemd distro if u hate it so much

18:45

linux gives u the freedom to choose

If any of you havent enabled magic sysrq, you really should btw

18:45

Got me out of this pickle

Ryozuki

linux gives u the freedom to choose

Not anymore with a majority of us deciding it's okay to let systemd be a monopoly

u still have the freedom

18:47

u just choose not to use it because u are lazy

18:47

windows users are lazy

18:47

become one with stallman

Barely have a choice. The only options with any sanity at all are devuan and gentoo

18:48

Any other I tried say they support sysv. But if you install discord or spotify it pulls in systemd lol

@Learath2 lets create our own distro

18:48

and make a init system in rust

18:48

monkalaugh

I would love to

18:49

Rust is an excellent choice for PID1

Teenux

18:49

the name

18:49

monkaS

For PID1 you want minimal and very safe

18:49

Pid1 goes wrong you get a broken system or a kernel panic

monkaS

teenux > systemd

poggers

learath2 > lennart poettering (edited)

18:55

hype

18:59

my keyboard doesn't have a sysrq key

18:59

it's the print key, right?

alt-print, yes

timakro

learath2 > lennart poettering (edited)

I'm not nearly cunning enough or have marketing skills like the guy to build a cult around my product

My new btrfs based automatical backup script does the unthinkable: I just pull the usb disk, no unmounting, call me insane

and honestly I think what he does is immoral so I wouldn't really be able to do it even if I had the skills

timakro

My new btrfs based automatical backup script does the unthinkable: I just pull the usb disk, no unmounting, call me insane

isn't this how you get broken backups?

how does the script pull the plug?

oh and that

I pull the plug :D

19:05

I decided all those filesystem and kernel engineers are putting so much thought into making fail safe (e.g. power outage) filesystems

why not just unmount if it's already idle?

someone got to counter that, right?

they can do their very best but anything that's being written while you pull it is at best suspect

and discover all these juicy bugs (with their data on the line)

I honestly would at the very least sync before pulling it

It's doing a backup every hour from different machines so I really can't be sure. I wouldn't wanna have to check or boot my pc to unmount

do you have the script running on a router or sth?

Btrfs is doing all this stuff (I assume ext4 too) to make sure everything is flushed before the final reference is moved in the b-tree or whatever

19:08

on a raspberry pi

yeah but how do you figure out if a backup is going on rn?

not ext4

wdym? I don't, I just pull the disk when I' swapping it out

btrfs or ext4, which one u use?

19:08

i use ext4

btrfs, I'm slowly moving my machines to btrfs

I'm confused, why would you swap the disk out?

ext4 only has protection of metadata

i saw btrfs have a regression in speed in the latests kernels

19:09

due to bug

I always want to have one not at my place, I'm paranoid like that :D

I don't see how you would not risk atleast incomplete data

https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.10-Btrfs-Regression

Yeah saw that on HN frontpage

Making rounds this week was word of a "500~2000% performance regression" knocking Btrfs on Linux 5.10.

19:10

watch nasa rover

If you are plugging the disk out you HAVE to make sure it's atleast not being written to rn, btrfs' approach would only guarantee it's safe for you to unplug it while a backup isn't going on

Image attachment

19:10

poggers

@Learath2 only if the backp isn't written with power outage in mind

@Learath2 I don't think so, yeah think of power outage like heinrich said. Nobody will every slap a "we guarantee this or that" label on it but in practice it just works

uuuh, you'll atleast lose that backup

yeah ofc

Please don't be lazy and wire a led to the raspi

I'm cleaning that up the next time round

if the led is lit backup is going on

I thought about that, maybe with a button :D But I was lazy

@timakro any reason besides it being in-tree to choose btrfs over zfs?

a button to stop the backup and unmount/sync would also be very pretty

@heinrich5991 I haven't really looked at zfs

zfs is also very cute

19:13

my old nas was running zfs

19:13

(on freebsd ofc)

my nas and teehistorian storage is zfs

@heinrich5991 on linux?

yes

The features I want are atomic snapshots and the ability to transfer them incrementally over the network like with btrfs send/receive

how is the support for it nowadays?

previously on raspi

19:14

yes, that plus content checksumming is also in ZFS

iirc zfs snapshots are also atomic

19:15

tho I don't use them so don't remember

@Learath2 not sure, it's still out-of-tree, but it's okay with archzfs

I considered linux on that NAS, then I remembered I don't really need anything from linux so might aswell be safe with freebsd

It's awesome that my windows virtual machine is backed up automatically now, also still boots when the backup was taken when it was running. And it's actually feasible now to do this. I had a rsync based solution before and you can only backup the entire image then, not only the parts that changed

19:16

@heinrich5991 Why did you choose zfs over btrfs?

because it looked maturer at the time, especially the raid stuff

19:17

more mature? idk, english adjectives are weird

I think they still say on the btrfs wiki that some raid levels are not working or "half working" ^^

more mature sounds correct to me

"more mature"

19:17

yeah

at the very least "maturer" sounds verywrong

both are right, apparently(?)

TIL

yeah dict.cc says both are right

19:18

never heard maturer

btrfs also does content checksumming, right?

I guess it sounds odd because of the ərə at the end

quitted exist

19:20

too

quitted? wtf

yes

19:20

Image attachment

19:20

but its use is gone

@heinrich5991 I don't know

that's also something I definitely want

There is checksumming but that might be only metadata?

19:22

@heinrich5991 Just checked, data checksumming is on by default

nice

So the tricky thing about my strange backup solution was ironically to unmount the disk and also the mapper device for the luks encryption ^^

so btrfs is zfs but made by linux aligned ppl with cool modern tech

19:23

if linus approves it then its gud

Because even when you plugged out the disk it's still mounted, and to unmount it you still need to kill all processes which are writing to it, they are still writing even if the disk is unplugged ...

19:24

So I've got a udev hook to do this, but udev hooks can't mount or unmount stuff because they are in their own mount namespace

you shold be able to join different mount namespaces, I think

And the popular solution seems to be to write a systemd service which does the mounting and call that from udev to escape the namespace

Ryozuki

if linus approves it then its gud

they almost cancelled linus a year or so ago

ohhh right, that's why i decided against zfs

19:25

it's in linux?

19:25

or not?

19:25

idk it was weird

yes

19:25

it's "in-tree"

19:26

i.e. in the linux source code

zfs?

19:26

is in the kernel?

no, btrfs

yeah

I wonder if it's a licensing issue or sth with zfs

19:29

how it's still not in tree is insane to me

licensing

19:29

it's a GPL-incompatible GPL-like license from oracle (edited)

ah oracle with their nasty influence over the open source world

19:30

I should move to btrfs to not support oracle stealing from the open source community

Also, you know when you are trying to unmount a disk and it doesn't work. You probably still have a shell open or smth which has it's cwd in there. There's really no way to reliably find out what's stopping the unmounting. There is like a huge list of things which can be the cause and nobody really has a comprehensive list of everything

ah actually it's the poisonous license of linux that's creating the issue here

19:31

hm, not sure which party I hate more

you mean GPL is poisonous?

I don't think either license is at fault, they're really similar

7

A few days ago I switched all my stuff from gpl to mit

it's just that they both require freedoms to be retained in derivatives, so they're incompatible

GPL probably has some clause that doesn't even allow linking to non GPL code

because they require slightly different stuff

timakro

you mean GPL is poisonous?

I guess infectious is a better word

19:34

If any part of your code is GPL all your code needs to be GPL in most cases

19:34

I think LGPL is the less anal version

You might need LGPL or something. I had GPLv3 on everything until someone who wanted to integrated my project into his complained and I realized I don't understand GPL good enough to use it

I still maintain BSD, MIT, zlib and the likes are the best choice for people that actually want their project to be used by the most amount of developers

I believe there is a point for GPL for huge projects like the kernel. With MIT, in theory someone could fork it, add a super duper feature everyone wants and distribute it without source

I wonder if GPL proponents disagree on this @Learath2

@heinrich5991 ofc they do, most people love to shove their ideologies down everyones throats

oh come on, this isn't an objective argument (edited)

Well don't they? Wouldn't you want everyone to agree with you on everything if possible?

19:37

If a person didn't think they were right they wouldn't argue their position in the first place

maybe their objective isn't "wanting the biggest amount of developers to use their project"

19:38

but instead exclude those who they think are working against their objectives

AGPL-3 best

19:38

poison the world with freedom (edited)

timakro

I believe there is a point for GPL for huge projects like the kernel. With MIT, in theory someone could fork it, add a super duper feature everyone wants and distribute it without source

Such is the nature of open source, it is well within your right to want more than credit for your project. I just believe it's a tad against the principle of open source as I define it

bluekitty

heinrich5991

but instead exclude those who they think are working against their objectives

Sure it guarantees a different set of freedoms, a set that I think restricts the spread of open source code

it guarantees user freedoms

19:40

i personally think all knowledge should be free and open

19:40

these science paper websites behind paywall are cancerous

@Learath2 it's not clear-cut without data. I'd be fine with being served some data about that

Ryozuki

it guarantees user freedoms

Sure, but (very subjective opinion here) I think it's mostly a secondary effect. It actually guarantees that people get to exclude people that don't think like themselves

and they dont even give a living to scientists

(due to possible second-level effects)

Learath2

Sure, but (very subjective opinion here) I think it's mostly a secondary effect. It actually guarantees that people get to exclude people that don't think like themselves

it guarantees to exclude the ability to remove those freedoms

It's how I believe the world works, I think people are actually extremely selfish

19:41

That's why capitalism had a very easy time permeating society

we had this convo before

we have differing welfare systems across the world

fuck the ddnet dev channel should move from discord to something proper

wdym

it's currently on IRC and discord

19:42

what's your suggestion?

19:42

(and by extension kinda also matrix)

I don't really have a suggestion

19:43

But as long as there is a bridge chances are I'm too lazy and using discord (edited)

19:43

Not only me

same xd

heinrich5991

we have differing welfare systems across the world

Sure, but even in the most socialist societies among us people participate in the society if it benefits them if it makes them happy

I don't see this working without a bridge without excluding lots of people (that would currently be here)

Look at all the people moving from france to monaco to save their precious tax bux

I could actually join via the matrix bridge

19:44

Didn't know we have that

19:45

I use element anyways

we don't, but matrix has a freenode bridge

19:45

and #ddnet also exists on freenode

I would love to avoid discord aswell, but it's too late. Discord took over the world

will be gone in ten years, don't worry

Another monopoly we all allowed

yeah

heinrich5991

will be gone in ten years, don't worry

No way

its sad ppl should use decentralized social media more

@Learath2 long-term bet? ^^

mastodon etc

The NSA would personally fund discord to keep it alive, it's a massive data farm

heinrich5991

@Learath2 long-term bet? ^^

Sure, 50e?

xd

19:46

bet a crypto

0.0001 btc?

50e in the future will be worthless

monkalaugh

if something is more volatile than euro its bitcoin

when corona ends

Centralized systems are here to stay, if anything there will be more of them in the future

massivle inflation

19:46

everywhere

It's extremely threatening to countries all around the world that people get to freely talk about what they want

@Learath2 we had lots of centralized systems that failed

19:47

so ten years from now, discord is smaller than it is today @Learath2? that fine?

Sure but what replaced them? More centralized systems

yes

19:48

I don't doubt that

actually about capitalism, it may look like it havent "failed" but its cuz u are kinda in the middle to good side of it

19:48

you are not getting the consequences of it

I'm just saying that discord isn't here in ten years

as much

heinrich5991

so ten years from now, discord is smaller than it is today @Learath2? that fine?

Well sure, I think I'll take that bet. Even with it's dozens of missing features people use it, as they improve the product more would use it imo

By June 2020, the company reported it had 100 million active users each month.[26]

Ryozuki

actually about capitalism, it may look like it havent "failed" but its cuz u are kinda in the middle to good side of it

The middle actually got the shaft with capitalism btw. Especially in societies that produce a lot

The freenode bridge is only for the dev channel?

brb eat

yes, that's correct timakro

Haha, I like that. ddnet is the dev channel

Well the only people likely to use irc are the devs, so it makes sense

@Learath2 so in january 2031 less than 100 million monthly active users

19:51

eh february

Sure. What is your hypothesis anyway?

my hypothesis is: centralized services come and go

19:51

they're not here forever

But why would it go away without reason? It's users adore discord

same was true for skype

do you have past examples for that heinrich5991 ?

myspace

Actually tab completing mentions works

19:52

awesome

Mh, I don't really see how one would improve upon discord. I always saw how one would improve upon skype

19:52

I think we are at a point in technology that the improvements are no longer obvious and large

Right, myspace. I also wouldn't be surprised if facebook was loosing users

Same as with other fields of science. Try finding sth new in math or quantum physics

losing*

I can't see youtube going away ever

youtube will live as long as google supports it

If you wanna heinrich xDD

Especially with the barrier of entry being INCREDIBLY large

19:54

Can you host 100 million users? I can't imagine anyone except the tech giants could take a task like that on

Well, you can grow

Maybe another corporation that promises to be more closely aligned to governments could be funded by them

19:54

I honestly don't see how google even let discord happen tbf

19:55

It would have been sooooooo easy for them to snatch this position from them

Do they have competition?

google failed like 5 social networks and fails providing a messenger

Problem with communication applications is that you need people to communicate with

19:55

It's easy to start your own communication application to "compete" with discord sure

@Learath2 legislation could break that btw, e.g. eu legislation

What is the point if my friends aren't on that platform? Same with whatsapp absolute monopoly in turkey

(e.g. by requiring interoperability)

They tried to kill it with both a govt funded app and Telegram lately with the privacy policy changes

19:56

Neither could succeed because if you can't send a message to your mother, no one wants Telegram

I don't agree that telegram is better than whatsapp

19:57

telegram does (for all intents and purposes) not have e2e encryption

The privacy policy change on whatsapp was bad enough that proper news channels suggested Telegram in turkey btw

heinrich5991

telegram does (for all intents and purposes) not have e2e encryption

huh, really? I thought they relied on the exact same tech as whatsapp. Is it opt-in or sth?

ouch, telegram's marketing is really nasty

19:58

that people keep believing this

19:58

telegram does not have end-to-end encryption outside of never-enabled badly working single-device smartphone-only "secret chats"

19:58

oh, and they're one-to-one only

Ah, wow that's scummy marketing

(also they roll their own crypto for client-server encryption)

heinrich5991

@Learath2 legislation could break that btw, e.g. eu legislation

They could try, but I feel it'd be very hard to get there. A lot of people oppose the government regulating markets. Most these companies get away with blatant monopolies for years with slaps on their wrists

20:01

Look at Facebook, it is no doubt a monopoly that keeps buying and buying other companies to grow their monopoly. Yet still no monopoly busting in the US or in the EU

I wonder if you could get funding which such an unoriginal idea as a centralized youtube clone

timakro: well there is dailymotion out there used by 3 people

Ofc, what we really want is a decentralized youtube clone but then the question is if the people funding you even know the difference

you would definitely get more funding for a centralized one than a decentralized one

im back

the old money people and big tech both hate the idea of decentralized systems as decentralization itself is against the idea of a corporation

i just looked it up and it seems like youtube and dailymotion started around the same time

20:04

so it's no surprise they both got funded (idk if they got funded)

(I actually tried to pitch a decentralized cloud photo storage application to some richppl before)

crypto is decentralized

did you have the word "blockchain" in your pitch?

timakro: DAMN that's what I was missing

then there's your mistake

actually the blockchain itself is a rly simple term

yeah it's two words

20:05

block

20:05

and chain

a chain of blocks

super simple

They suggested I ditch the E2E encryption and the decentralized aspect btw

highly effective in business meetings

that contains the top hash of the merkle tree of the previous block

20:05

iirc

how did it work Learath?

@Learath2 do u know rich ppl

20:07

@Learath2 timakro[m] have u heard of web3?

i'm genuinely interested in decentralized designs. it seems like 100x more difficult to do so 100x longer until you get profitable

https://web3.foundation/

Still better chances of being successful than a youtube clone which doesn't have anything to offer over youtube

20:08

imho

20:08

Nope, I havne't

Well my friend who pitched it with me pivoted into a centralized non encrypted version of the idea and got his funding. I bowed out because I didn't find it interesting

ew

(and I suffer from impostor syndrome so I get really insecure when things like this happen)

hmm

20:09

@Learath2 just think that most programmers just know html js and css and think js is the best lang in the world

20:09

after thatu will feel better

solid advice :D

20:10

So it built onto something from this web3 organisation?

20:10

Or was it a pitch to web3?

well a foundation is born form something that exists i think

I was asking learath about his project

ah

20:12

xd

Ryozuki-: I don't think the unique stuff is licensed, is it?

https://www.youtube.com/watch?v=2ocykBzWDiM&feature=emb_title

Rick Astley - Never Gonna Give You Up (Remastered 4K 60 FPS)

20:13

they remastered

20:13

rick roll

20:13

to 4k 60 fps

20:13

humanity at its best

20:13

Remastered music video with Topaz Video Enhance AI and RIFE (Flowframes).

yeah by remastered they mean slap a ugly upscale filter on

it looks good

Ryozuki

@Learath2 just think that most programmers just know html js and css and think js is the best lang in the world

You can't logic away irrational feelings. I do know this

No?

20:14

It looks like the latest cell shading game from nintendo

??

It's also the reason I get very angry with people suffering from dunning kruger syndrome

gotta search taht term

People feeling so confident with their knowledge with their portfolio consisting of a single hello world program

The Dunning–Kruger effect is a hypothetical cognitive bias stating that people with low ability at a task overestimate their ability.

20:15

ah like the russian guy

20:15

ye its very annoying

the russian guy was doing interesting things. please keep the personal stuff down

I feel like I'm inadequate for everything with my 11-12 years and here is this goof with his 1 week of experience thinking he is Knuth himself

this chat is all personal stuff since ever

okay, let me rephrase, don't put other people down in this chat

ur just biased heavily towards me

20:17

the guy doing interesting things "aka some hover effect" gives him freedom to talk shit about others?

no

it indeed does

What russian guy?

timakro Bananasth on github

someone with a cyrillic nick on discord

he is obsessed in telling me i dont know how to code

20:18

several times

Oh I remember some fighting a while ago -.-

point me to it when it happens (if I indeed miss it), I'll try to moderate that too

Well both you and him have personalities ranging from condescending and mean to verycondescending and verymean. So I would have probably muted you both if I cared about such behaviour

timakro[m] why the m?

matrix

Why does everybody under that youtube video say it looks good

20:20

I something wrong with me?

begging for approval

20:20

no, I also think e.g. the hair looked weird

is reddit down

maybe depends on the resolution you look at

20:21

at low resolutions it seemed clearer than the original to me

1440p

can't read the rust subreddit

20:21

sadge

The original looks low res, this looks just weirdly stylized

https://jemma.dev/blog/gc-mark-and-sweep

Ruby Garbage Collection Deep Dive: Tri-Color Mark and Sweep

In the first post in the Ruby Garbage Collection Deep Dive series, we went through a few definitions to give us a picture of how Ruby stores values in memory. If you haven’t read it yet, read it first! We’ll build on those definitions in this post. Particularly, we’ll talk more about the Ruby Heap, Pages, Slots and RVALUES.

20:22

Tri-Color Mark and Sweep

20:22

they always come with these names

20:22

funny

20:23

ruby syntax looks so weird

Maybe if they throw machine learning at it

OMG I found the next big thing

20:23

Neural Network GC

Image attachment

20:24

https://skyline.github.com

Your GitHub story in 3D - GitHub Skyline

View a 3D model of your GitHub contribution graph. Share it, print it, and more!

20:24

omg

I mean the upscale xD

i found the newest github fancy site

But neural network GC sounds nice

with synthwave music

Maybe if we add blockchain

20:27

Ryozuki-: Nooo your's bigger than mine :)

monkalaugh

20:27

if u have a 3d printer

20:27

u can print it

That's so dumb

timakro[m] r u watching nasa

no

:(

But seriously, why don't we move the ddnet discord to matrix?

20:33

It seems you can make "communities" there, it's like a discord server, you tried it.

20:33

just tried it

20:33

We can just go to matrix.org, no need to host anything

matrix communitiesi s just 1 channel iirc

because the players are on discord

[freenode] timakro[m]

But neural network GC sounds nice

Sounds awful, imagine your gc making a mistake and removing a used object

hey dev i need help :3

@WhiteKitty what's wrong?

@deen what i make wrong with that/

That doesn't look like official DDNet server. What is it?

vanilla

and it looks like a mistake in the source code, not from your side

21:41

vanilla has city mod?

yep

and is called Server-Open? I don't think so

21:42

Where did you get it?

i changed it in bam

21:42

from friends

21:42

but wht is wrong deen?

That doesn't work as easily as you think

21:42

or you copied the wrong file. Maybe you mixed up character.cpp and chatcmds.cpp

huh

I don't think you can just copy some files into vanilla to make it into a city mod

ye

21:48

@deen idk what i make with it xd

Sorry, can't really help with that.

done i fixed

21:51

xd

21:52

problem was on gamecontext.h i add wrong file (edited)

[ddnet/ddnet] Pull request opened: #3630 /times: Show country rank is from

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...

Exported 458 message(s)