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 2018-12-06 00:00:00Z and 2018-12-07 00:00:00Z

just leave the debug flag out?

[quakenet] <Learath2> yeah just cmake .. -G"MinGW Makefiles"

or type=Releaase?

00:01

works aswell

00:01

maybe my vs is old

00:01

idk

00:01

and some ddnet update killed it

[quakenet] <Learath2> I'd guess deens player sorting commit killed it if anything

how di i reset ma git to a specifiy commti agian?

[quakenet] <Learath2> git checkout commitsha

ty

00:06

ima rebuilding prior to deens sorting commit to ztest it

00:07

Seems like i am still using mingw how do i switch back to vs?

[quakenet] <Learath2> my modern c++-fu is pretty weak but from what I read I think it's about violating strict weak ordering

ah didnt deleted the cache

00:08

what is c++-fu ? xd

[quakenet] <Learath2> ChillerDragon: no idea, I haven't used vs to compile something in 5 years now

yy deleating the cahe did the job

00:09

then it used vs agian

00:09

y this one works

00:09

so any commit after a1db76cdeff4232dd0865dfcbf7250aef66a7a38

00:09

@deen pls fix

[quakenet] <Learath2> ChillerDragon: were you joining a ddnet server or a vanilla server with it when it crashed?

kog

00:15

based on ddnet 9x as far as i know

00:15

9.3.1

[quakenet] <Learath2> ChillerDragon: can you check line 2642 in algorithm?

where do i find alogrithm

[quakenet] <Learath2> it gives you the path to the file in the crash

Embedded image

00:18

looks like wrong indentation for me

00:19

xd

[quakenet] <Learath2> gotta love the cryptic code in these files :P

00:20

[quakenet] <Learath2> I wonder wtf DEBUG_LT_PRED even expands to

00:22

[quakenet] <Learath2> ChillerDragon: can you find out what that is defined as?

haha?

00:22

no u

00:23

its the first time that i even opened some of these libs i think

00:24

i have no idea where to search

[ddnet/ddnet] Pull request opened: #1391 Add command 'repeat' to repeat the text last one said

strange VS thing, never had any issue on windows when cross compiling master

bam server_release [1/1] #1 link DDNet-Server.exe link: extra operand ‘pthread.lib’ Try 'link --help' for more information. bam: 'DDNet-Server.exe' error 256 bam: error: a build step failed what is this?

bam?

00:34

isnt bam depreciated

00:34

we upgraded to cmake

00:34

well "upgraded" hehe

old src ( 10.6.2 ) (edited)

hm idk

00:36

what src

00:37

gosh google spams me with advertisements on youtube wtf

00:37

like google the search engine

00:37

they are dominant af why would they need ads

end of the year

00:39

left over money to spend

hm mby

00:39

but im sure they coudl still spend it more wisly

00:39

spend it on ddoisng duckduckgo xd

00:40

I would maybe switch to google if gugl takes ddg down with ddos but not if i see stupid ads

what did they advertise

Searching stuff is fun

00:40

some fortnite shit idk

[ddnet/ddnet] Pull request closed: #1391 Add command 'repeat' to repeat the text last one said

oops

google is affiliated to fortnite?

i made this guy rq

00:41

who is sirius even

00:41

some chinese guy

00:41

idk baut fortnite and what google does

00:41

it was this fortinite dance and a grandma

00:41

and some searching food or what ever stuff

00:42

2 different gogole ads i had in the last hour

they are dominant af why would they need ads

00:42

this logic is flawed btw

why

00:42

imo ads are usefull if people have to know your product

00:43

nobody knows that there are diffferent search engines

because you don't just stop when you're doing good

00:43

always pushing further

ads != doing good

00:43

no bing user will switch to google on that ad xd

Adds already pushed too hard today most people use addblocker because of that ^^

who needs adblocker

00:44

i have ddnet discord

00:44

xd

00:44

Rhytm best adblocker

don't think their target are different users

idk why i even use ma browser currently

00:44

lemme fix

they also profit from their users using it more

ah true i remeber

00:44

because its broken

00:44

@jao fix it

00:44

or i have to watch ads

what browser

#voice-text

00:45

Embedded image

00:45

there it is agian

00:45

and omg yt forced fokin darkmode header

looks funny

00:46

see they want you to search something specific

google will kill us all lol

00:46

they gave some ai access to their servers

what was that website

00:46

google will eat himself

00:47

something like that

idk it

http://www.gwei.org/index.php

conspiracy

00:48

@ChillerDragon do u wanna fix a cheat

depends

on?

idk tell me

00:48

what cheat

00:49

u should fix rythm

only in private

k

00:50

202.345.117 Years until GWEI fully owns Google.

00:50

xd

[ddnet-maps:master] 1 new commit

9c71c06 A Pizza - def-

@ChillerDragon that error you see is probably my fault with the new sorting I added :/

[ddnet:strict-weak-ordering] 1 new commit

0ad472d Make cort comparator conform to strict weak ordering - def-

06:58

[ddnet/ddnet] New branch created: strict-weak-ordering

06:58

[ddnet/ddnet] Pull request opened: #1392 Make cort comparator conform to strict weak ordering

comp(p1, p2) == false was violated when both are nullptr https://en.cppreference.com/w/cpp/named_req/Compare

Please take a look @ChillerDragon ^

@deen its fine!

@heinrich5991 do you have a functional teehistorian file or me :O? mine seem either corrupted or I don't know x)

@jxsl13 heinrich is busy :c

anyone got one off ddnet's teehistorian files for me? just trying to play around with it in python

11:28

of*

11:28

ddrace's*

11:32

@ChillerDragon why is he busy :O?

11:32

x)?

idk buys dude

11:37

i wait for his return since days xd and i notice people miss him all over the place xd

of, that kind of busy.

11:43

thought like a few hours not online busy

(edited)

11:48

maybe @deen or someone else has got a proper teehistorian file :O?

Can't you create one your self?

@jxsl13 i have a few hundred GB for you

12:24

what file size of teehistorian file do you want?

oh meh @deen u didnt do it on master :c

12:26

can you tell me the cmd agian to checjout your por

nope, only want to push if that's the fix

[ddnet/ddnet] Pull request closed: #1392 Make cort comparator conform to strict weak ordering

12:26

[ddnet:master] 2 new commits

0ad472d Make cort comparator conform to strict weak ordering - def- 19840bc Merge pull request #1392 from ddnet/strict-weak-ordering - def-

now it's on master, easier for me

oder so

12:26

ty

@jxsl13 there you go, 3 random ones from GER with different file sizes: https://ddnet.tw/teehistorian/

data leaks o.O

it's less than you can gather by spectating on a random server

12:29

doesn't even contain chats

yy ikr

12:30

btw your pr works c:

great

no crash when joining kog servers

12:30

ty for the fast fix @deen ❤

sorry about that, I didn't think about what requirements the comparator of std::sort has :/

12:30

and gcc didn't assert them

nobody could expect that

well, it does make sense

12:31

if a < b, then b > a etc

12:34

If anyone else wants to play around with teehistorian, @heinrich5991 has a nice Rust library here: https://github.com/heinrich5991/libtw2/tree/master/teehistorian/src

heinrich5991/libtw2

Some Teeworlds stuff in Rust. Contribute to heinrich5991/libtw2 development by creating an account on GitHub.

@deen what is the differnce between DDNET.ico and DDNet.icns in other/icons ?

12:51

i am trying to get a custom icon for my own client

open them ico is for windows, icns for mac

i opend ico on windows and tried to edit it with paint

12:57

no good result

don't use paint, use gimp

12:57

an ico has different layers for different sizes

hm dont have it on my windows i think i try some webconverter

there might also be a way to create it with imagemagick or other tools

hmm

12:59

it doesnt update

12:59

do i have to change something else

12:59

delete some cache?

remove the build directory and try again?

13:03

dunno

y thats what i am doing rn

13:03

lets see

13:04

yay it worked c:

13:05

Embedded image

13:05

love it

13:06

ty deen

I wanna read the format using Python and learn some data analysis vor whatever :D

13:10

vor*

four

or

[ddnet/ddnet] Issue opened: #1393 Server icon doesn't look like server icon

I just realized that the server icon does look pretty speed racy but not server at all. I don't know if this is an known issue because it seems pretty obvious. But an update would be nice c; maybe something like vanilla with a terminal.

btw, thanks, @deen

13:19

and wtf, 405 mb xD is that one year worth of data?

You're welcome

13:20

New teehistorian file gets created when map is changed

13:20

Servers get rebooted after 24 hours once they are empty

24 hours Oo?

That might just have been a popular map on a pretty full server for a day or so

man, that's damn large

we pack them after 1 week to save some space

how much space does compressing them save? to what size can you shrink that one file :D?

give me a few minutes to run xz -9

13:23

ah, but we only run xz -0f

too heavy on the server?

and doesn't save much more psace

hm k

found another 431 MB file, compressed with xz -0f it's 158 MB

13:24

@heinrich5991 took special care to make the format well compressible I think

hm, that's pretty good

[02:15:35] <Learath2> teehistorian@main:~/data/ger$ gzip -l 4193a96e-b0e8-4d1d-a8fb-edcf292e6245.teehistorian.gz
[02:15:37] <Learath2>          compressed        uncompressed  ratio uncompressed_name
[02:15:39] <Learath2>           161259316           481104542  66.5% 4193a96e-b0e8-4d1d-a8fb-edcf292e6245.teehistorian

this was one test I ran couple months ago

13:35

bz2 and xz get better results but also they are slower

[quakenet] <deen> xz -0 is faster and smalle than gz

13:37

[quakenet] <deen> and if you want really fast then use zstd I guess

13:37

[quakenet] <deen> really small I mean

I really should be compressing my teehistorian archive :/

13:37

I have like 781GB now

[quakenet] <deen> There is no good reason to use bz anymore, except for stone-age systems. xz is about 10 times faster to decompress and also compresses at similar speed but better compression

13:39

[quakenet] <deen> Don't compress the files newer than 1 week though

13:39

[quakenet] <deen> or they'll get recompressed

13:39

[quakenet] <deen> find data -name "*.teehistorian" -mtime +8 -exec ./archive.sh {} ;

13:39

[quakenet] <deen> retransmitted*

the internal UUID should correspond to th e file name, I suppose?

[quakenet] <deen> no

in the teehistorian file

[quakenet] <deen> yes

13:45

[quakenet] <deen> we use that to store the UUID of a rank in the database

13:46

[quakenet] <deen> and then we can look up the teehistorian file

now again, UUID , first 16 byte should correspond to file name?

[quakenet] <deen> Learath2: are you syncing ger2 btw?

@deen yes if you have my teehistorian ssh key in ger2

[quakenet] <deen> I guess

13:49

[quakenet] <deen> https://raw.githubusercontent.com/facebook/zstd/dev/doc/images/DCspeed5.png

Embedded image

@jxsl13 ?

[quakenet] <deen> lzma is xz, zlib is close to gz

the yes, no part kinda confused me, so the first 16 bytes are the uuid and are identical to the filename?

[quakenet] <deen> reminds me of the opus graph btw: http://opus-codec.org/comparison/

Comparison – Opus Codec

Opus, the open standard, high quality codec. Presentation, documentation, comparison with other formats, download links, source code repository.

13:52

[quakenet] <Learath2> my stupid bank keeps locking my card for "fraudulent" transactions, it's just me trying to withdraw cash from an ATM...

@jxsl13 afaik the filename is the entire uuid, and the uuid is 16bytes, so I think the answer is yes

[quakenet] <deen> 278 GB for me btw with xz -0 on everything > 7 days

where the heck do you store the files of th epast years?

[quakenet] <deen> Learath2: you might need to clean out your .ssh/known_hosts for ger2 though

We should all be archiving the same way so we can rsync all our backups together

13:54

@jxsl13 I have 781GB of them on my server

[quakenet] <deen> is xz -0 deterministic though?

13:54

[quakenet] <deen> might depend on version etc

13:54

[quakenet] <deen> I store them half on DDNet.tw and half on GER :D

13:54

[quakenet] <deen> but no "past years", teehistorian is not taht old

@deen yeah noticed known_hosts it should be syncing up now

any kinda magic number?

magic number of what?

I get the same value for all of these teehistorian files within the first 16 bytes xD wtf

hmm the first 16bytes might not be the uuid then

13:57

gimme a sec

13:57

I think it's a json header

heinrich said yes

13:58

https://openproject.johnbehm.de/projects/zcatch/work_packages/49/activity?query_props=%7B%22f%22%3A%5B%7B%22n%22%3A%22type%22%2C%22o%22%3A%22%3D%22%2C%22v%22%3A%221%22%7D%5D%2C%22t%22%3A%22updated_at%3Adesc%22%7D

13:58

got a quoted answer x)

https://github.com/ddnet/ddnet/blob/master/src/game/server/teehistorian.cpp#L63

DDraceNetwork, a mod of Teeworlds. Contribute to ddnet/ddnet development by creating an account on GitHub.

13:59

Yes it is uuid

13:59

Idk how you are getting the same thing then

jupyter notebook in a docker container x).. who knows

13:59

what might happen there

14:01

what's the encoding ._.?

The first 16 bytes should be just raw bytes of the uuid

14:03

heinrich5991 was adamant that endianness doesn't matter in this case

uuid is 32 values? but 16 bytes ._.?

14:04

hm

14:04

didn't say anything, 2 values are one byte

[quakenet] <Learath2> why does endianness not matter though, I still don't get it :P

maybe the uuid generation is symetric in that regard?

OHHH

14:09

I remembered now

14:09

The first 16bytes is the "magic" uuid

xD?

14:09

is it one sam euuids xD?

14:10

uuid?

// "699db17b-8efb-34ff-b1d8-da6f60c15dd1"

yes XD

14:10

damn it

14:10

699db17b-8efb-34ff-b1d8-da6f60c15dd1

static const char TEEHISTORIAN_NAME[] = "teehistorian@ddnet.tw"; static const CUuid TEEHISTORIAN_UUID = CalculateUuid(TEEHISTORIAN_NAME);

14:10

The actual uuid is inside the json header that follows the first 16bytes

14:11

and the endianness doesn't matter because you know the magic bytes

so there is a magic numbe,r eh

A magic UUID to be exact

next problem is, that after the zero-terminated header should be an integer, supposedly 4? bytes that define the message id and how to parse it

14:13

hm, those numbers should be negative that indicate messages, all I get is crap :c

Well did you read in the header?

yes

14:14

Image attachment

14:15

oh boi, this is exactly the wrong version x)

well just output 4 raw bytes from the file to see what's going on I guess

Image attachment

again, I'd just output 4 raw bytes to check it out

b'A\x1bF\x01'

14:26

65 27 70 1

hmm, that does look wrong :/

14:33

I wonder how CPacker packs ints

14:34

@jxsl13 yeah as I thought, it's not that simple

great xD

First bit is the extend bit, the second bit is the sign

14:35

the rest is data

14:35

That is for each byte

hm, k

Actually only the first one has the sign bit

wait

14:36

4 bytes

14:37

only the first bit of all of these should be then the sign?

When decoding the integer, you go byte by byte

14:38

The first bit of every byte specifies whether another byte is following it

14:38

The second bit of the first byte specifies the sign

dear jesus .. and thanks for the explanation

I think the last byte is also padded if needed :/

14:44

So basically just take 6 bits from the first byte 7 bits from all the others following it, if the first bit isn't set at any byte you break, at the end apply the sign according to the first bytes second bit

14:44

I'm going out to the bank, need to tear the branch manager a new one

i had a little idea today. when you kill yourself ingame i think it would be fun if a little animation plays where the tee makes the > < eyes, then shoots himself. any keypress interrupts the animation and the suicide so it would aslo work as a little kill protection

sounds like an intrusive feature

i forgot to mention it should be optional

that sounds like the kind of feature that u would use once or twice and never again

y

like telegun?

15:39

rekt

Lol

[ddnet/ddnet] Issue opened: #1394 Missing newline on whisper

I realized atleast on windows the client console looks really ugly if some whispering is going on because it fails to newline: !ss 2018-12-06 at 07 30 20

I added a new cpp file under game/client/components and a responding header file and i included the header in gameclient

23:45

but it seems like the cpp file isnt compiled because i have a lot of code in tehre that shouldnt compile but it still successfully compiles with 0 warnings

[quakenet] <Learath2> Did you add it to CMakeLists ?

no

23:48

iu guess i should

23:48

lemme have a look at the file

23:49

ah i see evertyhign is added manually thgere

23:51

ok nice ty @Learath2 i have my expected errors now c:

[quakenet] <Learath2> it used to be a glob, but apparently this is a better practice

Exported 331 message(s)