Teeworlds - bridge

Teeworlds

IRC / bridge

One-way IRC channel bridge. If you want to be able to send messages to IRC, contact @Dune or @heinrich5991. https://www.teeworlds.com/?page=docs&wiki=rules/irc_rules

Between 2020-01-11 00:00:00Z and 2020-01-12 00:00:00Z

I wonder why OSX is the only build that fails on const vec4 (&aColors)[] = {HealthColor, ArmorColor, AmmoColor};

what the heck is that syntax

^

(&aColors)[]

03:21

this part especially

it's an array of references

10:59

I thought that'd be C++ compliant and my gcc seems to agree

stackoverflow says no arrays of references Oo

12:21

it seems

12:23

the parenthesis' kinda look weird there ._., confused me.

You don't want a reference on an array

Can I thread a function in ListDirectory?

12:51

Storage()->ListDirectory(IStorage::TYPE_ALL, Path, TestCallback, this);

12:51

in this case TestCallback?

Oy: hm, getting LNK errors when I git clone teeworlds && bam :/

13:34

I think that's related to recent changes?

13:35

(win7)

Dune: both clean repos?

well teeworlds is freshly cloned

13:37

what's the second repo, bam?

i'll try

13:40

x64 or x86?

13:40

cl?

13:40

what link errors do u get?

x64 cl

13:41

FT & SDL

ok trying that

bam is on master

everything ok for me

13:44

could you do a bam config?

13:44

and paste the output?

sure

13:45

ah, that's it

13:45

bam finds mingw somehow

13:45

forgot about that issue I get with fresh repos

13:45

sorry for bugging :)

np :)

13:45

should probably do a gcc x64 check

oh, sounds good

should work now

I'll try

14:20

works! cheers

:)

o/

hey

cheers

hello

I'm currently working on multiple joystick support. Do you know whether I should save SDL_JoystickInstanceID or the UUID?

UUID seems to be SDL2 specific

14:30

not sure if it matters, aren't both unique?

instance id looks unique over restarting teeworlds to me

14:31

UUID seems more fool proof

Ah

14:31

Yeah

14:31

The GUID is build by squishing the bus, vendor, product and version numbers into a single value according to https://askubuntu.com/a/368711/843955

How to identify game controller for SDL2 in Ubuntu?

I'm trying to make Costume Quest (game by Double Fine) use either of my gamepads and according to developer's support all I need to do is add SDL2 config line to SDLGamepad.config file, however, I ...

14:32

Should be UUID to me

14:32

Oy ^

I'll change it to use the UUID and add some caching for the active joystick

14:43

I'll do a pull request later today

14:43

And get some food now first

day: forgot, you have multiple joysticks?

yeah should be the joystick guid when u wanna save settings etc.

Robyt3 is on a spree

Are there any features or plans for multiple rcon passwords or some sort of individual authentication to track which admin is acting?

pls give more space for gametype string xD

Hey, there need to be handicaps for this game

There is support for moderators. For the average server this should be enough I think

what kinda handicaps?

18:06

and why

like skill disparity handicaps

propose the removal of dyn cam

because the number one experience for a new player is to come in, get smashed by someone they can't touch, and then ragequit and move on to other games with more mechanics of fairness.

18:07

basically if your score outstrips the other players, it seems like your HP should go down or your weapon effectiveness etc

true, that's for modders and server hosters to regulate I'd say ._.

eh, I think that there should be some baseline default

18:07

"this makes the game playable for noobs too"

there are skill level icons

18:08

but well, dunno if anyone benefits from them in any way

yeah but noobs don't know those and that's not an actual solution, just a human ranking thing

18:08

yeah I have seen them

18:08

contextually, I've known teeworlds exists for like a month now

Due to the low number of players all skill levels play on the same server anyway

yeah

as there is no login system and whatnot to track players, it's kinda hard to prevent people from playing on specific players.

basically it should take less hits to kill advanced player, or their hits should do less damage, or both

18:09

and, the way to track this is at the level of in-game score

18:10

if somebody's score gets to be 2x everybody else's? adjust their HP or damage down

18:10

or even, maybe some fun mechanic like "the higher your score is the less bullets your guns hold"

18:11

make it possible to still win but definitely even the playing field a bit

I'm trying to make zCatch accessible for beginners, but it's far from perfect. it prevents some long time players from playing on the beginner server, but due to too few players the server gets rarely full.

18:12

changing gameplay mechanics to create a disadvantage for pro players is kinda unfair

as a noob zcatch is infuriating

18:12

the whole game for me is seeing if I can make the advanced players die sometimes when they're killing me

18:12

yes jxt that's the point

18:13

it needs to be unfair towards them because it's unfair towards everybody else because of them

18:13

they're preventing there being other pro players

that's why the beginner server exists, it tries to prevent players with a score of 35 or higher to play there, but as there are no accounts and so on, people could bypass the prevention

yeah but that mechanism 1) either doesn't work outright or 2) is insufficient

18:14

and as already pointed out, the low numbers of people playing means everybody goes to one server anyway

hm

18:14

uff, dune playing destiny 2

I kind of like the idea of shot restriction, down to maybe 3 rounds or something for somebody 30 kills ahead of everybody else

I'm thinking about regulating health

yeah health is the standard

18:15

but there's a lot of options

instead of anything else

18:15

but how would one determine a pro player

well the appeal for me is, some people have shit defense but good offense? and vice versa

18:16

so you could optimize to their level on each of these dimensions

sounds easy, not xD

well the basic determination is in-game score disparity

18:16

somebody getting way ahead

18:16

because that's just an integer the server tracks anyway

there are a lot of 0.5.2 players coming back, at least to zcatch that have a score of 0

uuuh k

maybe kd?

it's not the worst idea but the in-game-score thing works as well on noobs and anonymized players as everyone else

some people don't accumulate a score, like nameless tee

yeah, but you have an ID for the player on the server that's using that name

18:18

and you, haven't grokked the code but theoretically, you can aggregate the amount of damage done by each player?

18:18

and lower the damage done by the most-dealing player and up the damage done by the least

18:19

eh that also sounds like it could be frustrating, imagine being in last place because you get everybody down to 1 heart but someone else always kills them kind of thing

18:20

but, there needs to be something to address this

18:20

this is a thing

well, if damage is accumulated

18:20

it would not matter?

18:20

but it's not used for scoring, tho

yeah but that's the idea, use the data there to regulate game fairness

sound like a bad idea, better find an idea for a training system

18:27

zcatch is frustrating because it's a survival gametype, you have to wait before spawning, that's why instagib is popular for newcomers

18:31

hm, is there an flag for tees not to test tees collision ? I was thinking about ghost tees, like in race games (trackmania/mario kart)

a general one, should be tunings. player collision

rand no there's a real problem of the guys who can play at some certain level, making everybody else ragequit, and this really needs to be fixed

JewZeus: yeah zCatch is terrible if you're not in the top %

not just zcatch, but all gametypes

I think zcatch is particularly punitive

there's gametypes that the whole game is to be better at navigating than you can get playing the vanilla game

18:49

zcatch is yeah

but yeah all PvP gametypes are probably not fun if you're at the bottom of the bucket

the problem is 'the bottom' is the bottom 80% of the bucket

ctf is a bit like that by design

18:50

if you are worse than the flagger there is no way you can do shit to it because he'll just go faster and spam grenades back

eh that speeds up getting to the next map

18:50

but a bunch of servers freeze it on one map

18:50

so that doesn't actually work

almost all do

18:51

that's why most players play TDM in the middle in ctf5

18:51

because you can't do shit if the flagger is a better player (and it tends to be the best player of the server, cause others die with the flag)

eh there's no single flagger I've ever seen

18:52

unless one player is far and away an example of needing to have some kind of adjustment to his outputs in the game, for the sake of everyone else, as already mentioned

maybe not just one, but certainly some of the better players

right

@jxsl13 doesn’t the new wide browser give more space?

20:45

Regarding

pls give more space for gametype string xD

it doesn't seem that it gives any more space to gametypes

20:46

maybe half an ascii character wise

Hm I don’t know I just looked at map names

20:49

Was super upset I could not differentiate ChillerDragores20-29 maps because it was chopped of

20:49

But the wider browser solved it

don't change the strings of popular gametypes it just confuses and fractionates

Error copying directory from "/Users/john/Desktop/gamer/other/sdl/mac/lib64/SDL2.framework" to "pack_teeworlds-0.7.3-osx_dmg/teeworlds.app/Contents/Frameworks/SDL2.framework".

21:13

rip?

21:13

always the same problems with client

is /Users/john/Desktop/gamer/other/sdl/mac/lib64/SDL2.framework a directory?

21:18

oh, it says gamer

21:19

I think Dune doesn't keep CMakeLists.txt up to date for his mod

I do

21:19

unless something recent broke it

ok, sorry

I'm ig, will check later

/Desktop/gamer/other/sdl

21:23

every folder below doesn't exist

that seems to be the problem then, I guess? ^^

it has most likely nothing todo with gamer

cmake or the make_release script?

cmake

cmake

package or package_default target?

default

https://github.com/teeworlds/teeworlds/commit/cbb57765b66477e2a6daf6347e3fd7e9f366ce83#diff-af3b638bc2a3e6c650974192a53c7291

I was c&p from the github actions

cherry-pick that

hm

21:26

nice, thx

21:26

original tw gives me: Error copying directory from "data/maps" to "pack_teeworlds-0.7.4-osx_dmg/teeworlds_srv.app/Contents/Resources/data/maps".

21:26

even tho submodules are cloned

21:27

hm

21:27

maps are not being copied into my build fodler

21:27

folder

21:27

most likely

21:28

usually one creates a build folder in cmake in order not to build in source

21:29

data/maps -> datasrc/maps ?

21:30

hm, datasrc doesn't exist in the build folder

you're not building in the source folder, right?

yeah, there is a copy action missing somewhere to copy the maps into the build/data/. folder

21:32

no, I'm building in the build fodler

21:32

folder

I don't think there's a copy action misssing

are you usually building in source?

no

21:32

no one is, that's why it's untested

ok, good

it doesn't work afaik

didn't work either anyway

21:33

it tried to copy from data/maps

21:33

moving datasrc/maps to build/data/maps worked for me

your logs are unfortunately quite incomplete, so it's hard to tell what went wrong

21:33

sure

21:33

the code to copy the maps over is there

21:34

have you seen a warning in the initial cmake output?

let me retry from scratch

21:34

cd ..

21:34

xD

21:36

Dune, could you add this to your gamer repo? https://github.com/teeworlds/teeworlds/commit/cbb57765b66477e2a6daf6347e3fd7e9f366ce83#diff-af3b638bc2a3e6c650974192a53c7291

21:38

lol, now getting a linker error

21:39

oh, yeah oy's commit

21:39

it's a template netban that is ._.

21:40

if I read it correctly

21:40

bool CNetBan::IsBannable<NETADDR>(NETADDR const*)", referenced from:
      CServer::ProcessClientPacket(CNetChunk*) in server.cpp.o

21:42

hm, now it work

21:43

hm I think I know what's the problem, might have been my fault, cuz I did not use cmake .. again after initializing the submodules

@jxsl13 that's from 22 days ago, I'm surprised it's not in there

Can I thread a function in ListDirectory?Storage()->ListDirectory(IStorage::TYPE_ALL, Path, TestCallback, this);in this case TestCallback?

hm, gamer build seems to work fine now

21:50

after that patch

21:51

I added a pullrequest

what do you mean, thread?

i have a folder with 500+ files

do you want to start each TestCallback in its own thread, or do you want a thread for all these files?

laggs quite a bit if i listdir it

21:51

One for all ofc

ehm, the dmg package still says teeworlds 0.7.3 .-.

hm. one listdir shouldn't™️ lag I think (edited)

is that intended?

Heinrich, 500+ files

21:52

still increasing

21:52

So, can I thread the whole thing? just one thread to loop throught all files?

@Sonix server full :P

yes, you can start a thread and call this function in it

how?

21:53

i tried, but couldnt get anything working

what is the result of your directory listing?

i have some data in the files, and they will be saved in a vector

ddnet/0.7?

0.7

Oy sry didnt notce

21:55

come now

I think there might be a lot of cmake commit missing, I'm not sure

btw: another issue with listdir, i tried to run listdir with 1000+ files, and it crashed at some Point, but that is a different thing, the thread is more important now

21:56

So heinrich, how do i thread the listdir

either use the job interface of teeworlds, or spawn your own thread

Whats better?

I don't really know tbh

if you want to know when it's done better use the job interface... just easier

on ddnet I patched the job interface to be slightly more usable, but that hasn't made it into vanilla yet

hm, I might be stupid xD

21:59

gamer client on gamer branch

but okay, let's do the job interface

Yes

create a function that you'll call in the new thread

oh damn... 0.7 still uses the thread_sleep stuff

yes

@heinrich5991 wdym

it'll need to have the signature int Name(void *pUser)

Ah wait, you mean thread_init_and_detach(ListDir) and then in ListDir I call the ListDirectory function?

22:01

(looked at file_score)

yes, but I thought you wanted the job interface

Oh

22:02

i dont even know what that is tbj

22:02

tbh

file_score is not the best source when you want to see proper threading code ^^

True

22:03

@heinrich5991 int function created

you probably want to pass it some data, right?

Yea

22:04

Can it have access to gamecontext?

then create a struct that contains the data you want to pass it

22:04

no

22:04

that would make it thread-unsafe

Ah true

22:04

Mh but it would need it :)

22:04

:/

why?

22:05

what data does it need?

uhm

22:07

idk how to explain

22:07

Well

22:07

can I make it read only?

no

std::future

22:07

xD

no?

22:07

But I can pass the data?

you need to make a copy of your data that you want the thread to know

Ok.

22:08

That could work

.. I do some threading stuff in my mod ..

you create a struct with the data that the thread needs

but not the c way

And what if the thread needs a function?

22:08

Also to the struct?

put function pointer in struct

22:08

._.

I definitely need a function in it

22:09

Ahh this is complicated rn

gamer's cmakelists says: file GLOB requires a glob expression after CONFIGURE_DEPENDS.CMake (file)

Gotta eat something, heinrich are you there in 20 minutes?

yes

wow... the jobpool is only used for dns lookup

and sound loading IIRC

22:18

yup

and hostname lookup for favourites

ah okay

22:19

what about using semaphore here like ddnet does?

why does no one uses semaphores?

sounds like a good idea to not burn cpu cycles

the array of references error persists, I removed the parenthesis and it properly said that

22:23

error: 'aColors' declared as array of references of type 'const vec4 &' (aka 'const vector4_base<float> &'

you said above that arrays of references are forbidden, right?

22:26

that should be the error cause then

yeah

22:26

dune said that he intended to create an array of references

22:26

and that it build on gcc

22:26

._.

apparently, it's just disallowed by the standard

22:27

it may work in gcc, but that doesn't mean anything

yeah, that's what SO said

i.e. it needs to be replaced by something else

1337 * o/

I'm just gonna replace it with the array without the references

22:29

it's a vec4 that's being passed by value

probably uses that array to modify it?

22:29

so it might need to be a pointer

it's a const array

22:29

and the function that is being passed elements from that array takes actually no reverences, but the object class vec4

22:29

all values are const within that array

hm ok

and the array ist const

22:30

is*

heinrich5991: about the version server. could you try to run the 0.7.3.1 one?

22:32

to see if that one worked

does anyone know where the sdl2 version requirement of the tw client is specified?

22:36

tells me tw requires version 11.0.0 and I have 1.0.0

@heinrich5991 im back

there shouldn't be a requirement

@heinrich5991 interesting that the standard doesn't allow that :)

22:37

yeah you can just remove the reference

interesting that clang > gcc >:D (edited)

apparently it is due to array access decaying to pointer arithmetic and pointers to references not being allowed

ooh, neat

client dies at runtime telling me "fu"

22:39

first it doesn't find th sdl2.framework, dying and then dies after I downloaded and placed the sdl2.framework file into the tw app/Frameworks etc

22:39

same fun as last time

u r on mac?

ye

@heinrich5991 appearently, i could also make this thing different without using listdir

22:40

i guess

do u compile it urself?

Oy: 0.7.3.1 versionsrv running

trying to compile the gamer client

try to compile master first

heinrich5991: thx

tw master or your master, dune?

same

I did compile the tw master just fine

I don't see why the mod would be different :/

and it even starts

unless I broke the cmakefiles

I opened a pr, that was giving me errors

@jxsl13 maybe check the git diff of the CMakeLists.txt

I'll just enable github actions and troubleshoot it on my own :)

I added the tw master as remote origin, kinda merged i into dune's gamer branch

22:44

and I see no cmakelists diff ._.

22:44

what's weird

doesn't sound too good, the original cmakefiles doesn't work?

with the pr I opened it works

22:45

is also builds

seems that one doesn't work either :(

and without?

lol

@heinrich5991 i have run ListDirectory now with 1112 files, and there is this error: Run-Time Check Failure #2 - Stack around the variable 'aUsername' was corrupted.

one message of me was not sent

and without?

discord seems to be having problems today

is also builds <- red

and without?

discord seems to be having problems today

22:46

e.g. that message didn't get through

@heinrich5991 ive run listdirectory now with 1112 files, and there is this crash: Run-Time Check Failure #2 - Stack around the variable 'aFileName' was corrupted.

the cmake .. step already doesn't work

you have an out of bounds write somewhere

22:47

try running it in valgrind

ah lol

22:47

i just

probably the aFileName variable

the cmake .. step already doesn't work

increased the

@jxsl13 yeah it's on me, it seems cmake broke at some point...

i just increased the char array size, now it worked. because i just copied files

test

22:48

wow

so the name got longer - Kopie

probably the aFileName variable

discord bad

i just increased the char array size, now it worked. because i just copied files

every second message works

it builds with bam on OSX

probably the aFileName variable

so the name got longer - Kopie

come on IRC ;)

we can switch to IRC if you want

i just increased the char array size, now it worked. because i just copied files

we can switch to IRC if you want

._. irc

22:49

old but works

22:49

old but works

https://webchat.quakenet.org/?channels=teeworlds

heinrich5991: the version server u build, is it x86 or x64?

64 bit

so cmake stopped working at some point with my cmakesfile on my mod; that one is on me. I never tested it on OSX, might be why

22:50

bam+OSX works

ok

bam bad

22:51

._.

22:51

is ok

I gladly accept cmake fixes ;)

I dunno, would dropping the tw one with your added files work x)?

probably, I did nothing fancy

22:55

ah, only cmake generates artifacts? :(

ddnet only compiles with cmake nowadays, it works quite well

22:56

or has done so for over a year

bam also uses less strict compiler options

22:57

the workflow job for bam is rather minimalistic

22:57

but it was the same before

accidentally typed matricks instead of matrix while writing gtihub's build.yaml :D

^^

this is love

23:15

x)

23:15

I swear to god, that globbing stuff with assets and not source files is annoying ...

23:17

I don't see much either other than you commenting out globbing and adding your few files to the cmake.

assets aren't globbed either

23:18

it just doesn't fail the build if they don't exist

startig gamer from console works

23:21

starting

23:21

packaging fail ._.

probably the files aren't listed properly in cmakefiles for osx packaging

23:22

different binary name etc :/

i made bam produce artifacts

weird

23:26

works

23:26

xD

23:30

remove CONFIGURE_DEPENDS in those three lines

23:30

and then it will build

removing CONFIGURE_DEPENDS sounds like a sure way to break the build in unexplainable ways

23:32

where do you want to remove that?

it's at the end of a file(glob "..." here)

23:32

without a glob expression behind it, that's what I'm getting in vs code

23:33

file(GLOB ENGINE_CLIENT "src/engine/client/.cpp" "src/engine/client/.h" CONFIGURE_DEPENDS) file(GLOB GAME_CLIENT "src/game/client/.cpp" "src/game/client/.h" "src/game/client/components/.cpp" "src/game/client/components/.h" CONFIGURE_DEPENDS) file(GLOB GAME_EDITOR "src/game/editor/.cpp" "src/game/editor/.h" CONFIGURE_DEPENDS)

@jxsl13 https://github.com/Dune-jr/teeworlds/suites/395881445/artifacts/1071063

` for code blocks, then it will not do formatting

please don't, on bridge

for his one line

yeah, that wa slike the biggest I'd paste in here

ah sorry :)

oh, it was three lines

yeah

@jxsl13 if you remove CONFIGURE_DEPENDS, it will not recheck the source files when you add/remove some

Image attachment

23:35

is that intendex?

23:35

itended

23:35

intended!

23:35

icon dead, maybe

23:36

the file doesn't open

23:36

the app

which?

23:37

the one built by github?

yeah

23:38

you need to edit pinfolist

23:38

or how the file is called

23:38

and tell macos to open not teeworlds but teeworlds_gamer

23:39

"Info.plist"

23:39

is the file

23:39

xml file

23:39

after that

23:39

the program crashes on startup, seemingly

23:40

/Applications/Teeworlds 3.app/Contents/MacOS/../Frameworks/libSDL2-2.0.0.dylib (which was built for Mac OS X 10.15) Expected in: /usr/lib/libobjc.A.dylib

23:40

I highly suspect that I might not run it, because I'm on macOS 10.14

23:40

but I'm not sure

23:41

I don't like paying with fire, so I'm not updating for at least another year

23:41

x)

23:41

dyld: lazy symbol binding failed: Symbol not found: _objc_opt_respondsToSelector

23:44

I think the client would need like the server the ./teeworlds shutdown command

23:44

in order to see, whether it crashes after startup and closing it gracefully

23:50

http://www.mediafire.com/file/sxic29mpws5pmmr/teeworlds-gamer-0.7.4-osx.dmg.zip/file

23:50

this should work

23:50

at least it did on my machine

Exported 496 message(s)