Teeworlds

discord.gg/teeworlds / development

For discussions around the development of the official Teeworlds

Between 2022-02-07 00:00:00Z and 2022-02-08 00:00:00Z

bruh not a bridge again

00:02

just mixing two versions is ugly af

00:02

3 would be horrible

there you have your answer

then don't merge. move.

00:03

00:03

how do you move? repeating all the ddnet changes on the 0.8 base? that is absolutely gigantic

make an AI that does the job for u

well, I was being sarcastic... but yes... that or doing your own rendition of DDnet for 0.8 lol just DDnet rewritten with hindsight in mind

00:06

since I did hear that DDnet had some "bloat" problem? I don't know if that was recognized though

00:07

cause one thing is bloat perceived from the outside and bloat perceived from the inside

this is all hypothetical dev resources anyway

00:08

we have just enough dev resources to decorate the grave right now

certainly

00:14

at least @Robyt3 is doing good work

00:14

does the DDnet also suffer from a dev shortage?

00:17

cause if it does, it makes such outcome even less possible than it already is

Stiopa

and I do feel like the plan fell flat. Got a few "that's cool" but not much beyond that...

doesn't help oy has not said a word about it either so far ;/

00:22

don't even know if he is in the process of checking it out or anything...

that's too much work to test, something like this would need to be played out a decent bit and get feedback

well, I need any and all acknowledgements

00:24

we've had games, we've had some maps created

00:24

but the radio silence doesn't help on the other side

00:25

if people will see that there is an interest from the maintainer...

00:25

then getting interest of the players shouldn't be an issue

00:26

whereas currently it feels like some project that might not even see the light of day

00:27

The Draft PR has been there for almost half a year by now and the seeming radio silence does not help one bit honestly

00:28

even a simple "This would need to be played out sometime" would give a clear sign to people that oy would be interested in having it tested... somewhat

00:28

even if it might not end up anywhere, really

I've also been demotivated to work on Teeworlds at all. I planned to do the HTTP masterserver for Teeworlds first for the longest time. I think more maintainers that you could talk to are necessary. this is not an attempt to get into the maintainership position

That'd be great

00:31

Having to rely on a single person for resemblence of "official thought and future" is not ideal, especially if they are hard to reach

00:34

Doesn't help oy's response on the maintainer issue is just confirming that the low activity is reason to oy's inactivity...

Stiopa

then getting interest of the players shouldn't be an issue

I don't share that opinion

Well it is something I felt while talking to a few community members

00:41

They didn't really see... the future in the project... They liked it, but were skeptical of how it would fit into oy's vision of the game etc. whether there was sense of committing to it further

heinrich5991

I've also been demotivated to work on Teeworlds at all. I planned to do the HTTP masterserver for Teeworlds first for the longest time. I think more maintainers that you could talk to are necessary. this is not an attempt to get into the maintainership position

direction would help a lot indeed. any other maintainer would be good progress, unfortunately there are very few (other than you) that have the intersection of skills and knowledge that it would take to be one

I was able to get a semi-competetive game of 5v5s on earlier versions of the build and map, but as I went inactive and the "vanilla pro" community went into a decline, it has been getting harder

but even when maintainer feedback is good, this community is comparatively atrociously bad at giving positive feedback

haha.... I can somehow feel it

00:44

I know I ran a google document through people ^^

00:44

but well the visions themselves were not unified and some visions conflicted between people

00:45

didn't feel that bad though

00:45

of course there was much room for improvement

00:46

I felt like the way to get proper feedback on gameplay loop etc would be to just let random games run... and watch the demos

00:47

to see how people would tend to play more naturally?

00:53

But, umm, yeah, I feel like the conversation has slightly derailed

in terms of community here, they dont like changes so feedback was and is always a hassle be it a mod or a map

Dune

0.7 client mods only exist because of a small handful of people that aren't really active anymore
0.7 skins almost don't exist
there are a few 0.7 server mods, but it's mostly version bridges or worse ports of existing mods. the new mods have zero players at the moment, and ZOMB3 doesn't even have an open server anymore

i made a 0.7 client aswell as a 0.7 server mod, that bridges to ddnet, its the other way around :)

Can the automapper handle rotations?

Stiopa

does the DDnet also suffer from a dev shortage?

Yep, especially people capable of porting something the size of ddnet to 0.7. We only have a handful of people that have the skill set to undertake that but none of us really have the time nowadays

Stiopa

Can the automapper handle rotations?

no

I started several rewrites of ddnet, but teeworlds really isn't set up well for writing maintainable mods. I always end up having to spend hours of my time trying to rework the base to allow for clean modding. Then real life came up and I sort of gave up on ever finishing a full rework, so I just made the bridge instead

Dune

no

so does this part of the code just not function?

Image attachment

And I was just trying to port the server, the client patchset is also vast and some of the 0.7 changes are completely politically unacceptable to our players

okay, question... how do I link to a particular part of the code?

I looked in the code to answer you, looks like I missed a part of the code then

12:34

you can do that on github

click on line numbers, then press 'y' to update the URL in the URL bar to a permanent link

Dune

pretty sure it is technically infeasible to move their current diff from 0.6 to 0.8 with their current dev resources (in a reasonable timeframe) (edited)

^

Stiopa

okay, question... how do I link to a particular part of the code?

You can add #Lxyz to the end of the url to hilite line xyz

https://github.com/teeworlds/teeworlds/blob/1ef3771bf641fe2d98ecba5596bb4dba5c2a9473/src/game/editor/auto_map.cpp#L50

Image attachment

Learath2

And I was just trying to port the server, the client patchset is also vast and some of the 0.7 changes are completely politically unacceptable to our players

like the packets for the name changes?

idk how I missed that line, I was wrong

Dune

Click to see attachment 🖼️

hey just did that

I think it worked

uh, I didn't know this menu

Stiopa

like the packets for the name changes?

A lot of the things that moved out of snap are annoying for devs but we already figured out how to workaround those while developing the bridge. For the client the rendering changes, especially the way text looks seem to annoy people for some reason. For the server the patching out of weak hook is a big issue

Some maps rely on the weak/strong hook?

Yep, not only that we also have a leaderboard that can be harder/easier to beat if we change hook physics

Right I can see how that is a really major red flag for moving

Anyway, nothing that's impossible to workaround, it's mostly the fact that we'd need a lot of time figuring out tricks to workaround things like these, time we just don't seem to have :/

well regarding weak hook... hypothetical 0.8 DDnet client could render that with no stutters, right? it would only be the 'base 0.8' client that would have problems But yeah, rewriting everything up to that point would be... a struggle

the textrender would probably be one of the biggest struggles to port or unify

12:44

i tried merging master into the editor branch to get that somewhat up-to-date but I haven't looked at the textrender compile errors yet

amount of differences grows with each passing oy's PR merge session

though there are some PRs for DDnet that are put into the teeworlds repo too, right?

system is mostly the same for both

12:46

I have a growing list of minor commits to cherry-pick from one into the other

ShootXen

in terms of community here, they dont like changes so feedback was and is always a hassle be it a mod or a map

about resistance to change, it does feel like so. People got too used to the current status quo.

predictably so. 0.6 lasted seven years. that means almost every player hasn't seen the game changed since they arrived, when 0.7 was released

well 0.7 seems on a road to last even more...

well, given the reduced development activity compared to 0.6, it is going to take longer to reach a threshold where compatibility break and content reset is worthwhile

13:23

it's debatable 0.7 reached it after seven years

I bet teeworlds wouldn't have so much problems, if we would allow Clients to support older servers and Servers to allow older maps, too. Every major update is really compatibility breaking and adding a new divide. If 0.8 Clients support 0.7 Servers, too, there would be no need not to upgrade your Teeworlds Client

yeah that's been mentioned...

14:43

normally it wouldn't be a problem but with bulk of the community playing mods

14:43

it becomes a real issue

Assa

I bet teeworlds wouldn't have so much problems, if we would allow Clients to support older servers and Servers to allow older maps, too. Every major update is really compatibility breaking and adding a new divide. If 0.8 Clients support 0.7 Servers, too, there would be no need not to upgrade your Teeworlds Client

then those are 0.7.x clients. major updates are defined by compatibility break

then we should work on 0.7.6 :p

well... there is a 0.7.6 milestone

teeworlds hasn't introduced new features with backward-compatibility "hacks" so far

14:51

basically compatible for the most part, but some non-gamebreaking prediction errors somewhere

14:52

this is how ddnet stayed compatible with 0.6 while still introducing network changes similar to 0.7

they don't always to be "hacks", in the case of maps a simple converting tool would be sufficient

14:53

and yes, you can't convert everything obviously

Assa

then we should work on 0.7.6 :p

yes, that's a better plan than 0.8

Assa

then we should work on 0.7.6 :p

https://github.com/teeworlds/teeworlds/issues/2135 is a blocker for any next version IMO

heinrich5991

teeworlds hasn't introduced new features with backward-compatibility "hacks" so far

hasn't it? those have been plentiful in 0.7.x. client-side race, chat commands,

I remember implementing one for the added skin-changer

Robyt3

https://github.com/teeworlds/teeworlds/issues/2135 is a blocker for any next version IMO

why this in particular? this is a severe issue but the existence of these hasn't prevented us from updating teeworlds

Dune

hasn't it? those have been plentiful in 0.7.x. client-side race, chat commands,

fair, haven't thought about that

Dune

why this in particular? this is a severe issue but the existence of these hasn't prevented us from updating teeworlds

if you really want a release, sure, but this severely limits the server browser so it should really be solved

was this introduced in a commit after 0.7.5?

yeah, I think so

ah, that wasn't what I understood

14:58

my bad then

14:58

@oy oy added this to the 0.7.5 milestone on 12 Jan 2020

doesn't seem like it

that's just the issue I think

14:59

the commit where oy fixed it is only on master

I don't understand

14:59

the bug was introduced before 0.7.5

for me this is a real blocker, I would have fun adding a materials layer with all kind of different properties and physics, but I really have trouble with the map standard itself. Also this might be blocked by the new editor

so we released a version regardless of this bug persisting

14:59

so it shouldn't be a problem to release more versions with this severe bug?

ah, sorry, I meant together with the comment by heinrich in on that issue

15:00

because it got worse due to a commit is only on master yet

Assa

for me this is a real blocker, I would have fun adding a materials layer with all kind of different properties and physics, but I really have trouble with the map standard itself. Also this might be blocked by the new editor

why are these "blockers". why shouldn't we update teeworlds with a 0.7.x until everything we really want is in?

Robyt3

because it got worse due to a commit is only on master yet

ah, I see

so now we can't even fit 16 players into a server info packet

it's a blocker because I can't do it

Assa

it's a blocker because I can't do it

that's not what blocker means, it means you can't release before it is fixed (edited)

(not even 16 players in the worst case)

true

2790 would be amazing to get...

15:04

regarding https://github.com/teeworlds/teeworlds/issues/2135 it seems like a pressing issue to fix.. but there does not seem to be much of a discussion on how to fix it :/

one could fix it like ddnet did

15:06

(similar to, not exactly the same)

I briefly tried merging the ddnet-serverinfo but teeworlds had already added a different token implementation so I didn't know how to combine them

ah, I really wouldn't try to merge it, the ddnet one is a bigger hack due to the fact that 0.6 server info was so non-extensible

yeah, basically have to rewrite it

yes

heinrich5991

ah, I really wouldn't try to merge it, the ddnet one is a bigger hack due to the fact that 0.6 server info was so non-extensible

well it does seem like ddnet trying to keep up with the teeworlds 0.7 is far from ideal...? like, it seems like if not for the politically inexcusable changes, moving ddnet to a potentially more modern version would be quite the positive, barring the huge amount of effort needed to bring it to such a point

what does it bring to ddnet?

ddnet moved to an entirely different serverbrowser model

15:14

https://master1.ddnet.tw/ddnet/15/servers.json

15:14

all servers are in here

ohey, json

hey, you motivated me doing something again

(all 0.6 and 0.7 servers are in there, if someone wants to prototype a client implementation for 0.7 you can already get all the servers from there)

15:16

historical data is there: https://ddnet.tw/stats/master/

Assa

hey, you motivated me doing something again

That's why I like those discussions

can you tell me where I find the map-datatype?

15:20

my guess is map.h :p

15:21

what's the state of the new editor?

Image attachment

2

15:25

I see the numbers speak for themselves

https://github.com/heinrich5991/libtw2/blob/143dc845a67623002487f61b850168c929b04a45/doc/datafile.md https://github.com/heinrich5991/libtw2/blob/143dc845a67623002487f61b850168c929b04a45/doc/map.md

15:27

some map documentation

just saw the video with ice physics, I find it weird that you gain speed on ice cause there is less friction

15:29

real physics says less friction --> harder to accelerate

please don't use acceleration and speed as synonyms

15:30

the acceleration on ice is worse, it's 0.4 times the default

15:30

but your max-speed is higher, because there is less friction

15:31

imagine it as having a higher terminal velocity and it starts making sense

heinrich5991

https://github.com/heinrich5991/libtw2/blob/143dc845a67623002487f61b850168c929b04a45/doc/datafile.md https://github.com/heinrich5991/libtw2/blob/143dc845a67623002487f61b850168c929b04a45/doc/map.md

working through it ❤️

15:33

version_header: magic xD

https://en.wikipedia.org/wiki/Magic_number_(programming)#In_files

Magic number (programming)

In computer programming, the term magic number has multiple meanings. It could refer to one or more of the following: Unique values with unexplained meaning or multiple occurrences which could (preferably) be replaced with named constants A constant numerical or text value used to identify a file format or protocol; for files, see List of file...

what does a front-layer do in ddnet?

it's a separate layer that mostly accepts the same entities as the game layer

15:39

so you can stack game tiles

can you give me an example where it's usefull to stack game layer entities?

15:39

ah yes, a grenade inside a spawnpoint

Also rebase the Ice physics PR to resolve the conflicts

Assa

can you give me an example where it's usefull to stack game layer entities?

mostly used for hookthrough tiles

Saw the video about ice, velocity is damn too high when you run

15:44

But very cool, can be fun to play with

Robyt3

Also rebase the Ice physics PR to resolve the conflicts

do I really need to rebase or is an interactive pull enough?

ReiTW

Saw the video about ice, velocity is damn too high when you run

already tweeked max_speed, the video is outdated

15:46

I used the values of run_icicle, you can check it on the Unique race servers

Stiopa

Click to see attachment 🖼️

the downside of refactorings

rebase is cleaner than merging master. I never resolved conflicts using github's tool

is run_icicle only on the CAN server?

editor2 should be really separate files though

15:48

rebasing doesn't seem urgent to me

Dune

editor2 should be really separate files though

that's what you would hope, but we changed things that the editor2 interacts with

Dune

rebasing doesn't seem urgent to me

I meant the Ice PR, so it is actually in a mergable state

15:50

the text render was changed but the usages in editor2 were not adjusted, the rest is minor things like moving constants around that can just be solved with a text replace

nvm

15:50

i was looking at the short run server

Stiopa

is run_icicle only on the CAN server?

It's on every race server, but on middle, you can switch to the desired maps by using chat_commands: /map run_icicle (edited)

yeah I just tested it

that map works with a tune-layer and therefore you don't have the fancy animations

yeah it felt surprisingly smooth

@heinrich5991 what is stored in CMapItemLayerTilemap in int int m_aName[3], if the layers have names, we woudn't require flags

15:57

mapitems.h

the name as shown in the editor

16:00

you can rename layers

but you can't rename the game layer for example

16:01

so i guess for game it's game, right?

not via the editor, but the editor recognizes a renamed game layer

16:01

there are maps out there with localized names for game layer, at least russian and german

@heinrich5991 lol https://github.com/teeworlds/teeworlds/blob/6c6a3a0bb39caff5af643fc369c5edef2bbd3be9/src/game/editor/layer_game.cpp#L10 (edited)

16:14

I think they then use other clients

apparently

why not flag a layer as a game layer and load each gamelayer in a dynamic array

16:23

I mean why by default that was done like that, with the name

it's flagged and the name has no purpose

ah, never went to read that part tho, don't know how it works

but it would be smart to use names and I tell you why:

16:25

ddnet for example adds multiple other layers with their own flags like TELE_LAYER=2; and if we add other layers with their own flags now, too, we would break compatibility

16:26

therefore checking with name=="Teleporter" is smarter

But then it would require to set fixed names, as with flags you could just create a button "Create teleporter layer" smth like this

16:26

well you can do the same with the name but

the names can be fixed in the editor, your localization should only be used LOCALLLY

16:27

and not in the map itself

16:27

if I load a map a chinese player made, i don't want to have the gamelayer renamed in something chinese

Well still someone could name a tile layer "Teleporter" if nothing is done to stop it (edited)

Thats why I simply added another flag with the name OTHER and check for FLAG > 0

yeah that's a good way to go with

now we only have problems if people decide to implement the same things xD

Well for ddnet, they went too far from the original game, would it be really necessary to keep any compatibility with them instead of just going forward & implement stuff

16:31

(by far I mean there's many features like those tele layers etc..)

I was thinking the opposite, maybe we should add some features like teleporters and speedtiles

16:32

you can still disable them for pure mods

Well I was always comparing teeworlds with quake, teeworlds looks a bit like a 2d version of quake, with less guns

16:33

would be cool with teleporters yeah

16:33

Even quake had the hook

new features of ddnet use UUIDs

16:38

e.g. for the map, there's the automapper config that has an associated UUID

16:38

https://github.com/ddnet/ddnet/blob/0a7ff4f9d0e23e3ea7e6ad8c53204419bf79b5a6/src/game/mapitems_ex_types.h

16:38

(this would correspodn to the UUID 3e1b2716-178c-3978-9bd9-b11ae0410dd8, e.g.)

there are a few bugs with the datafile though

16:42

Image attachment

16:42

five years ago or so I decided that they're not exploitable. I hope I'll stay right

only the one I hotfixed was really exploitable

ah, I meant "not exploitable beyond crash"

and the latest snapshot fix might also have been, but not realistically

16:50

but there was actually an exploit map that would open the calculator

oh wow

but it probably doesn't work on every system

so it was exploitable. nvm then

https://mmmds.pl/fuzzing-map-parser-part-1-teeworlds/

16:51

gif doesn't work..

Additionally, the binary even has the stack protection disabled what makes it even easier.

16:52

that's an issue

16:52

Unfortunately, there were no other bugs that could be chained to bypass ASLR. For the purpose of the exercise, I continued with ASLR disabled and use ret2libc technique to pop a calc executing system("/usr/bin/xcalc").

16:53

I think that was part of the reasoning why I thought it's not exploitable, ASLR

ah, that makes sense

16:53

would be unlikely to be exploitable if the OS has ASLR

[teeworlds/teeworlds] Pull request opened: #3055 Text Selection and IME support!

Text Selection

!sel Implement keyboard-based text selection for console, chat and editboxes. Implement mouse-based selection for editboxes. Closes #12. Closes #1851. Word skipping (when holding Ctrl) is overhauled to be consistent with the Windows / Firefox experience that I took as reference. Password/Hidden input rendering and handling of unicode strings is fixed in the p...

Exported 245 message(s)