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 2022-05-21 00:00:00Z and 2022-05-22 00:00:00Z

a very stupid question, im just starting in coding and everything if my git bash is in c/users/user and i want to git clone ddnet, would the clone of ddnet be installed in that directory or do i have to specify a specific folder for it to go somewhere else? (edited)

Try it out?But yes git clone creates a folder in your current directory unless you specify a custom path using the second positional argument

Patiga

@Jupstar ✪ I'm truly embarrassed of my older self for not really having tested the colorspace thingy for webgl properly when I did it

i wouldn't blame you tho It's not very intuitive to only have sRGB framebuffers

06:25

according to GPUInfo every GPU that supports a sRGB format, also has a UNORM format and vice versa

Image attachment

[ddnet/ddnet] Pull request opened: #5177 Fix HUD rebase (thanks to bencie for report)

Checklist

[x] 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...

@c0d3d3v why did u split the ddnetcharacter into this new display thing?

09:04

like, why do we have two extended characters now xD

I tried to put it in ddnetcharacter but I could not get it, so old clients still can parse ddnetcharacter and new clients can parse the old ddnetcharacter ... maybe I did something wrong :/ (edited)

09:15

maybe we should not care at all?

09:15

and just expend and do not care about old versions?

fokkonaut

@c0d3d3v why did u split the ddnetcharacter into this new display thing?

have you experience with expanding net obejcts?

You can just add new fields

09:42

but u need to mark them i think

09:44

validate_size = false

09:44

then u can extend them

I also looked in the history a bit, and never saw a change of net object after a release...

ofc

09:44

GameInfoEx

09:44

just add that parameter and extend it

09:44

only to the end tho

09:45

not inbetween

09:45

iirc

but ddnetcharacter does not have the validate_size set to false... so old clients will fail to read it or not?

09:53

In theory it is also not very important that old clients can read it :/ it does not contain game relevant data

[ddnet/ddnet] Issue opened: #5178 Set log_level?

Right now everything is always logged, including tracing.

or wait validate_size only checks if the size of the received obj is to small o.O so normally it should just work on old clients. I will test it out. (edited)

you should always be able to add new elements at the end

I see

i think

I do not know what I did wrong, maybe I did add some stuff in the middle

:D

but in principle it is also not to costly to add a new net object... just 32bit more (obj id and snap id) as if you would add it to the end of an old obj. But I will try to move it to DDNETCHARACTER (edited)

its just not nice

10:05

thats why we always try to think of all usecases before adding a snapobject :D

[ddnet/ddnet] Pull request opened: #5179 Add loglevel config (fixes #5178, fixes #5174)

Checklist

[x] 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...

c0d3d3v

but in principle it is also not to costly to add a new net object... just 32bit more (obj id and snap id) as if you would add it to the end of an old obj. But I will try to move it to DDNETCHARACTER (edited)

+16bit for the size flield "_size is a field that is only present if the size of items of type type_id is not pre-agreed for the protocol. A list of pre-agreed item sizes can be found in the appendix of this document." xD

fokkonaut

thats why we always try to think of all usecases before adding a snapobject :D

-.- ya I will move it

u dont need freezetick btw

10:15

we decided to use freezeend

10:16

so that it doesnt get updated every snap

10:16

also, isinfreeze is useless

10:16

because u can take it from the map itself, but actually in terms of modding, yes keep it in

10:19

also, why u need rampvalue and target? those are available in the client already

10:20

By this, other clients can even find out the cursor position of others. not entirely wanted

[ddnet/ddnet] Issue opened: #5180 Don't use STUN servers for LAN

No use anyway. @heinrich5991

[ddnet/ddnet] Pull request opened: #5181 Fix pickups moving on their own (fixes #5137)

Checklist

[x] 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...

i just pulled git master but i think there is something wrong with prediction when i hook, hold a or d and open chat

fokkonaut

also, why u need rampvalue and target? those are available in the client already

I'm guessing due to some floating point mess target ends up getting predicted wrong, probably should bw fixed in the client rather than sent back

nori

i just pulled git master but i think there is something wrong with prediction when i hook, hold a or d and open chat

@c0d3d3v might be related to your HUD changes?

https://github.com/ddnet/ddnet/pull/5153

keep input if chat is closed by C0D3D3V · Pull Request #5153 · ddne...

fixes #4653 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 s...

11:17

maybe this

11:22

@nori if you have the time for it you could try to bisect the commit, but it’s probably indeed pr5153

boop

13:09

boop2

boop3

Boob4

You ruined it

it already started wrong

13:13

it should be boop0

13:13

we programmers here (edited)

justatest

13:13

yeee

beantyping

13:13

PepegaChatter

Murpi noob tho (edited)

13:14

Oh shit, sorry i shouldn't edit here pepeH

pepeH

I was hoping to fix this:

[2022-05-21 15:10:20][ERROR][bot]: Event 'on_message' caused an exception
Traceback (most recent call last):
  File "/home/teeworlds/pybot/ddnet-discordbot/.venv/lib/python3.9/site-packages/discord/client.py", line 343, in _run_event
    await coro(*args, **kwargs)
  File "/home/teeworlds/pybot/ddnet-discordbot/cogs/moderator.py", line 198, in on_message
    or is_staff(author) or author.bot or f'<@&{ROLE_MODERATOR}>' not in message.content:
  File "/home/teeworlds/pybot/ddnet-discordbot/cogs/moderator.py", line 29, in is_staff
    return any(r.id in (ROLE_ADMIN, ROLE_MODERATOR) for r in member.roles)
AttributeError: 'User' object has no attribute 'roles'

by adding author.bot to moderator.py, since webhook messages are just bot responses

13:14

eternal sadness

Ryozuki

we programmers here (edited)

I see i am not the only one who edits

ye

13:15

irc is ded anyway

U

murpi

I was hoping to fix this:

[2022-05-21 15:10:20][ERROR][bot]: Event 'on_message' caused an exception
Traceback (most recent call last):
  File "/home/teeworlds/pybot/ddnet-discordbot/.venv/lib/python3.9/site-packages/discord/client.py", line 343, in _run_event
    await coro(*args, **kwargs)
  File "/home/teeworlds/pybot/ddnet-discordbot/cogs/moderator.py", line 198, in on_message
    or is_staff(author) or author.bot or f'<@&{ROLE_MODERATOR}>' not in message.content:
  File "/home/teeworlds/pybot/ddnet-discordbot/cogs/moderator.py", line 29, in is_staff
    return any(r.id in (ROLE_ADMIN, ROLE_MODERATOR) for r in member.roles)
AttributeError: 'User' object has no attribute 'roles'

by adding author.bot to moderator.py, since webhook messages are just bot responses

weakly typed language problems smh

13:16

with rust u wouldnt get that

13:16

BASED

cammo

Ryozuki

BASED

well then rewrite the entirety of ddnet in rust troll

troll

(╯°□°）╯︵ ┻━┻

┬─┬ ノ( ゜-゜ノ)

I was rewriting a lot of this bot in rust actually, just hit a bit of a speed bump after I got annoyed at how meh async callbacks are in rust

14:23

After writing Box::Pin for the 500th time I got a brain aneurysm and stopped working on it, maybe with a fresh mind I can give it another go now

14:25

I've achieved peak rust: for<'f> fn(&'f ctx, &'f msg, &'f args) -> Pin<Box<dyn Future<Output=CommandResult> + Send + 'f>>

do u have the source somewhere

14:42

what u are doing sounds like async traits

14:42

https://docs.rs/async-trait/latest/async_trait/

async_trait - Rust

github crates-io docs-rs

14:43

@Learath2 https://smallcultfollowing.com/babysteps/blog/2019/10/26/async-fn-in-traits-are-hard/

14:45

async will improve when GATs finally get stabilized

14:45

hope its soon

14:45

its the most awaited feature

14:45

pun intended

14:46

https://github.com/rust-lang/rust/issues/44265

Tracking issue for generic associated types (GAT) · Issue #44265 ...

This is a tracking issue for generic associated types (rust-lang/rfcs#1598) Repository Detailed tracking information is available at the GAT initiative repository Unresolved questions: #87479 -- de...

14:46

Image attachment

14:46

we so close tho

Ryozuki

do u have the source somewhere

Unlikely, I never managed to get it to properly work in a nice generic extensible manner, so I deleted each of my attempts that didn’t end up working

14:47

I made a very rigid version so I could have a working bot for now

@Learath2 honestly trying to generalize a lot is often not worth

14:47

specially if u only end up using it

14:47

but thats in general

14:47

not just rust

14:47

i think

Here it’d be worth it imo, it’d make adding new features so much more pleasent

well if u upload it someday i would like to take aa look

14:48

its true that async is not the best in rust sadly

14:48

and the rust contributors try rly hard to get it right xd

Why is F1 logging everything now, even when console_output_level is on 0

14:49

thats not how its supposed to work xd

It is better than the async in c++ which noone seems to use at all

i never used async in c++ either

14:49

xd

fokkonaut

thats not how its supposed to work xd

It’s probably the console unification patch. I don’t remember how heinrich did it but the dbg_msg’s now getting routed to the console might not have a proper level

fokkonaut

thats not how its supposed to work xd

and why do we need date stamps in F1 :D takes lots of space

Learath2

It’s probably the console unification patch. I don’t remember how heinrich did it but the dbg_msg’s now getting routed to the console might not have a proper level

yea

The next part of that PR implementing log levels will probably fix this issue

dbg_msg through console sounds not correct at all, at least when using them before console got created

I think dbg_msgs ALSO get routed to the console, not ONLY through the console

ah

Might want some module filtering and level filtering and default those to emulate the old console

yep

Anyway, I’m happy that we don’t have to instruct people to open the black console on windows, and launch the game thru console in linux

xD

PepeA

@Learath2 we didnt need that before too

14:54

console_output_level 2

14:54

only dbg_msgs werent there

14:54

i think

We did, some information necessary while helping someone fix an issue is only printed on stdout

nori

i just pulled git master but i think there is something wrong with prediction when i hook, hold a or d and open chat

on what server?

15:16

mh I see

fokkonaut

also, why u need rampvalue and target? those are available in the client already

ramp value is not availible in client for other players

why?

15:20

U have their vel and tunings

fokkonaut

we decided to use freezeend

freeze end does not contain the information when freeze started

Why do you need that?

fokkonaut

U have their vel and tunings

how would you get there tunings?

15:20

for freeze bar

depending on their tunezone?

15:21

tunes are saved in the map and therefore available in the Client

only if you predict

15:21

prediction is off if you spectate

15:21

or cl_predict is off

Well, you can still code it that way to get the tunings? xD

why would i

because its useless to send it

fokkonaut

Why do you need that?

Freeze start and end are needed for a bar because you need to know the total size of the bar to fill in the correct amount

we are in 2022 fiber age

Learath2

Freeze start and end are needed for a bar because you need to know the total size of the bar to fill in the correct amount

yep

its only 16 bit * 64 max ... (edited)

I definitely don’t agree with shoving derivative things in there just because you don’t feel like computing it on the client

It’s fiber age, we can run the client on our servers and stream the graphics aswell

why did all of the last changes get merged so fast... I couldnt even take a look

it was online for 4 weeks

But definitely not the commit with ddnetcharacterdisplayinfo as an extension to the ddnetcharacter xd

it was

Lol

but does not matter

15:25

you can change it if you want (edited)

no, you should do it and that should happen before the next release

15:25

in order to minimize versions with this as people tend to not update

the client should not do the job of the server. the server sends the client data to display not the other way around

the server shouldnt send data twice

The display info should probably get merged into the ddnetcharacter. We should ask trml about the velramp. We should send the freezestart so we can use delta compression, not freezetick

freezetick is freezestart

bad name then

15:27

:p

was not my idea

c0d3d3v

the client should not do the job of the server. the server sends the client data to display not the other way around

On the contrary, the client HAS to re do some of the job of the server for things to be correctly predicted

xD

The way teeworlds has been structured relies on the client being able to accurately do the work of the server, which is why we share the gamecore

The display info should probably get merged into the ddnetcharacter

already started.... fokkonaut requested it (edited)

fakk off

15:28

yeee

Anyway, other than those small things, the new hud looks nice

Yes

15:29

true

fokkonaut

fakk off

???

he wrote fakkonaut xd

sus

Learath2

The way teeworlds has been structured relies on the client being able to accurately do the work of the server, which is why we share the gamecore

but it currently does not predict the world good enough, and I do not want to fix it, at least not now

But by sending it from the server you are permanently taking up space in the networking code. We’ll have to send these values to clients for a decade

we could just force cl_predict and do not allow turn of cl_predict.... so remove it, that would improve prediction a lot. Also do not turn of prediction in spectator

I think prediction really needs some work, some things should always be predicted no matter what

if we remove cl_predict entirely, I would be ok to not send the data

15:34

would not be prefect...

15:34

but better then nothing

@trml do you remember why we don’t have prediction in spectator mode?

I think he is in vacation...

why do u need pred in spec

15:37

oh wait

15:38

i always confuse pred with antiping

fokkonaut

also, why u need rampvalue and target? those are available in the client already

target is just the true target, angle is wrong, it is to rounded to be precise , I do not even know why teeworlds used the angle instead of target... -.- (edited)

c0d3d3v

on what server?

ddnet

c0d3d3v

target is just the true target, angle is wrong, it is to rounded to be precise , I do not even know why teeworlds used the angle instead of target... -.- (edited)

what do you use the target for?

to display the angle of tees and hookline of cause

15:39

hookline was wrong of other tees, now it is correct

I dont know if I like the target being sent

but we can keep it wrong.... most of this game is wrong anyway

nah

would also be nice to display the cursor of other tees in spectator (edited)

15:41

xD

15:41

maybe

15:41

or in demos

Maybe there is some usecase for cheaters if they see other cursors?

15:41

idk

in ddnet?

other gametypes

15:42

well yea i think its okay

mh

15:44

It is more precise then angle, but I think for bots the normal angle would be enough... so would only make them more accurate too xD (edited)

15:46

and i have not seen a mod that uses the cursor position for anything other than determining the angle. But of course it would be possible to do more with it.

my mod lets you place things on cursor and create portal on cursor

15:47

i use the cursor for quite some stuff

uuh

15:47

ok

anyways

15:47

is there some

15:47

method to enable/disable DDrace hud

client side

15:47

cl_ddrace_hud

serverside i mean

do not send ddnetcharacter xD

15:48

could add a flag

when does the client use the info?

15:48

yes, please add a GameInfoEx flag

15:48

my mod relies on the ddnetcharacter but also needs the hearts and shields from time to time

15:48

depending on which gamemode you are in, because you can also enter a damage based area basically

mh can you change gameinfoex after joining?

yes

mh k

its snap object

15:49

i think a flag for both, hearts/shields and ddracehud is good.

thought it is only send once

nope

fokkonaut

i think a flag for both, hearts/shields and ddracehud is good.

if one of them is set, put everything top left corner, if both are send, do ddrace under hereats/shields

15:50

if none is sent, do none

15:50

shouldnt be much code change, just adding another gameinfoflag to m_Flags2 of GameinfoEx

so 3 flags, heards shields amo? ddrace hud

how is the ammo displayed in ddrace hud? not at all?

yes

bruh :/

15:51

then definitely 3 flags, yes,

4?

1. hearts, shields 2. ammo 3. ddrace hud

xD

15:51

k

i think hearts and shields shouldnt be separated

15:52

but thats actually good, having ammo without shields/hearts, but above the ddrace hud

kk

its good to keep this kinda stuff, for modders

15:52

ty :)

[ddnet:master] 3 new commits

2818c6e Fix HUD rebase (thanks to bencie for report) - def- c4bcf3d Render full text when it still fits - def- c1919f2 Merge #5177 - bors[bot]

16:15

[ddnet/ddnet] Pull request closed: #5177 Fix HUD rebase (thanks to bencie for report)

nori

https://github.com/ddnet/ddnet/pull/5153

I really should not fix bugs so late xD thanks I wrote in PR correctly what I wanted to do, but have it programmed the wrong way around lol. And I forgot to test it the way you did it, I just ran straight, so I did not notice.

also can i disable the unfreeze timer bar under the tee? if not it would be great to have an option (edited)

16:18

or maybe make an option stars or bar (edited)

16:18

even better

cl_show_freeze_bars 0 (edited)

1

cl_show_freeze_bars or somthing like this... there is a option in settings (edited)

stars are unnecessary if there is bar

bencie

cl_show_freeze_bars 0 (edited)

yes it is thanks

:thonk:

is it matrix bridge?

16:24

or irc

[ddnet/ddnet] Pull request opened: #5182 Fix chat prediction

reported by nori https://cdn.discordapp.com/attachments/293493549758939136/977531792389177364/recording.mp4 I have somehow made the condition wrong xD although I had written correctly in the PR what it should do. also i noticed that teeworlds does not allow you to hold the hook while chatting, but we have allowed this since 2014....

Checklist

[x] Tested the change ingame
[x] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related ...

@nori could you test https://github.com/ddnet/ddnet/pull/5182 you can download the build from here https://github.com/ddnet/ddnet/actions/runs/2363673296 as soon as has finished

Fix chat prediction by C0D3D3V · Pull Request #5182 · ddnet/ddnet

reported by nori https://cdn.discordapp.com/attachments/293493549758939136/977531792389177364/recording.mp4 I have somehow made the condition wrong xD although I had written correctly in the PR wha...

ok testing

Image attachment

17:00

not a server bug, i cant see the name

17:00

in scoreboard the name is there, means the name should be available and valid

17:00

Image attachment

17:00

spec char also shows his name

name of what? are you in spec?

that tee has no name, but he should have one

17:01

sec

@c0d3d3v it's fixed

Image attachment

17:02

Was there some client changes?

17:02

im on 16.1

maybe part of deens name elipsses change

okay funny

17:04

when a player is in a minigame, his name is not shown

17:04

Image attachment

17:04

no idea why, was not like this before. has to be a client bug, as client has all infos

what do you send diffrent in mini game

not much tbh, wait

17:05

ohhh

17:05

i send a spec-char

17:05

thats it probably

xD

17:05

yep

spec-chars being there dont let the the main tee show his name anymore?!

probably

why that?!?!?!

17:06

i used it to indicate where another player would spawn again when he leaves a minigame...

have to look in the code

can you please fix it

17:07

https://github.com/ddnet/ddnet/blob/master/src/game/client/components/nameplates.cpp#L339-L350

ddnet/nameplates.cpp at master · ddnet/ddnet

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/nameplates.cpp at master · ddnet/ddnet

17:07

this is it

17:07

not else if !!!

Image attachment

https://github.com/ddnet/ddnet/commit/e65bfa63504ceee4e918e7128b1b124eb41c73f7

fix rendering of nameplates for spectating tees (bug introduced by ...

…870)

17:08

@c0d3d3v why did you do that change?

why not, who would have known that you use spec chars xD

Learath2

@trml do you remember why we don’t have prediction in spectator mode?

I think main reason is that the snapshot already contained all the things the client used to show before, so it's an optimization (and also because unpredicted positions are shown)

c0d3d3v

why not, who would have known that you use spec chars xD

well, its not an issue in general, and if a server sends a character aswell as a spec char, both should work independently and shouldnt be separated like this :D Especially not if not all cases get separated instead only the nameplate rendering xD

it is also clipped ...

if a tee is there, surely his nameplate should be rendered

if you have specchar the clip position is overwritten

17:09

we need to handle it speratly yes

https://github.com/ddnet/ddnet/commit/e65bfa63504ceee4e918e7128b1b124eb41c73f7

17:10

well

17:10

i dont see a single reason for this commit

17:11

before it was as it should :D https://github.com/ddnet/ddnet/commit/e65bfa63504ceee4e918e7128b1b124eb41c73f7#diff-2df247c37322e22203c91873e17f82645ef6bc22437460f87160a31e310a2027R341

17:11

this function takes care of the specchar renderpos already

in ddnet you can not have both

c0d3d3v

in ddnet you can not have both

that does not matter

fokkonaut

well, its not an issue in general, and if a server sends a character aswell as a spec char, both should work independently and shouldnt be separated like this :D Especially not if not all cases get separated instead only the nameplate rendering xD

.

17:12

@c0d3d3v could you please take a look again, and let both render side by side?

c0d3d3v

if you have specchar the clip position is overwritten

I already told you yes

ty

'-'

17:23

imo freeze stars always were kind of redundant (@nori)

stars are unnecessary if there is bar

it should be optional

17:24

not enabled by default

17:24

i still use stars but still should be optional

17:24

bcs no need

[ddnet/ddnet] Pull request opened: #5183 render allways the nameplate of a specchar (or the real player)

fokkonaut uses spec chars in addition to normal players in his mod, so the names should also be displayed in addition to the player name

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
[x] Considered possible null pointers and out of bounds array indexing
[x] Changed no physics that affect e...

so yes

@fokkonaut test it pls https://github.com/ddnet/ddnet/pull/5183

looks good

c0d3d3v

we could just force cl_predict and do not allow turn of cl_predict.... so remove it, that would improve prediction a lot. Also do not turn of prediction in spectator

that is possible, but from what I understand the problem is not lack of prediction but that prediction fills in useful values? that could be done by using the gameworld for everything, but not using it to predict ahead

murpi

imo freeze stars always were kind of redundant (@nori)

true xD because of the shields

if this is wanted I could perhaps use make a small pr

c0d3d3v

true xD because of the shields

with stars you can see others tho

originally I tried to keep the prediction stuff separated from rendering stuff, to make stuff not too dependent on each other (so the gameworld would write into some intermediate structs that the rendering would use), but not sure if that really was cleaner (as it required an additional layer of things)

hm

Murpi transformed into a bot?

trml

if this is wanted I could perhaps use make a small pr

It would have to predict all players of all teams, at every time (so show_all would also be forced), to make the display perfect. So I vote for NO!

17:37

also prediction in demos would be needed

i am noby

17:48

huh

c0d3d3v

It would have to predict all players of all teams, at every time (so show_all would also be forced), to make the display perfect. So I vote for NO!

yes, main reason not to would be optimization (also the reason prediction for other teams was removed)

its just not needed

[ddnet/ddnet-discordbot] Pull request opened: #32 Fix webhook spam

Logs: `[2022-05-21 19:25:10][ERROR][bot]: Event 'on_message' caused an exception Traceback (most recent call last): File "/home/teeworlds/pybot/ddnet-discordbot/.venv/lib/python3.9/site-packages/discord/client.py", line 343, in _run_event await coro(*args, **kwargs) File "/home/teeworlds/pybot/ddnet-discordbot/cogs/moderator.py", line 198, in on_message or is_staff(author) or f'' not in message.content or message.author.bot: File "/home/teeworlds/pybot/ddnet-discordbot/co...

some values like freeze_start could perhaps be inferred (without using prediction) if you look at the first tick it was received/updated and take the time difference, but that would perhaps require making some assumptions (like freezetime being a multiple of a second) and not be too pretty

if you cant interact with someone else u dont need to predict them

yes, this was only regarding the question of whether to use inferred/derived/predicted values or send everything over the network when rendering

a

trml

some values like freeze_start could perhaps be inferred (without using prediction) if you look at the first tick it was received/updated and take the time difference, but that would perhaps require making some assumptions (like freezetime being a multiple of a second) and not be too pretty

mh, do we do freeze bars for other tees? if so this wouldn't really work with tees that are clipped out when freeze start

17:54

s/$/s/

i think on 0.7 the ninja bar is made only using timeleft

fokkonaut

i think on 0.7 the ninja bar is made only using timeleft

Duration of ninja is set, by tuning iirc, no?

but then its updated every tick

17:55

eh

17:55

No

17:55

sec

if it's constant, even easier to do

yes, true, then it wouldnt

https://github.com/teeworlds/teeworlds/blob/26d24ec061d44e6084b2d77a9b8a0a48e354eba6/src/game/server/entities/character.cpp#L853

teeworlds/character.cpp at 26d24ec061d44e6084b2d77a9b8a0a48e354eba6...

A retro multiplayer shooter. Contribute to teeworlds/teeworlds development by creating an account on GitHub.

17:58

pCharacter->m_AmmoCount = m_FreezeTick + Config()->m_SvFreezeDelay * Server()->TickSpeed();

17:58

this is how i do 0.7 bar for freeze

so 0.7 sends ninja activation tick? and ddnet desnt send it yet?:P

freezetick gets set whenever the tee gets frozen in Freeze()

trml

so 0.7 sends ninja activation tick? and ddnet desnt send it yet?:P

yes, since they have a ninja bar

trml

so 0.7 sends ninja activation tick? and ddnet desnt send it yet?:P

actually they send the end tick

Image attachment

[ddnet:master] 3 new commits

2cb4758 fix condition of BugDDRaceInput, do only reset innputs on non ddnet mods - C0D3D3V 71669e3 release hook in mods that do not allow inputs while chatting - C0D3D3V 38c8b5b Merge #5182 - bors[bot]

21:09

[ddnet/ddnet] Pull request closed: #5182 Fix chat prediction

Ryozuki

Click to see attachment 🖼️

as a scratch user i 100% agree and disagree

chilledragon found you https://www.reddit.com/r/ProgrammerHumor/comments/uusg8k/this_is_how_i_win/

21:36

chillerdragon

Ryozuki

Click to see attachment 🖼️

where is GML in this list

[ddnet/ddnet] Issue opened: #5184 Swap proposition resets swap timer of previous swap proposition

Is this a bug or a feature?

Ryozuki

Click to see attachment 🖼️

rust isn't even listed?

Chairn

rust isn't even listed?

its the black on black logo in the bottom-left

[ddnet/ddnet] Pull request closed: #5181 Fix pickups moving on their own (fixes #5137)

22:41

[ddnet/ddnet] Pull request opened: #5185 Fix pickups moving on their own (fixes #5137)

Supersedes #5181

Checklist

[x] 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-ad...

Trying to initally build a custom client on Windows with Visual Studio DDNet client by right-clicking the DDNet project (not the solution) and select "Select as StartUp project". does it mean Set as Startup project not select as? or am i doing something wrong

Image attachment

[ddnet/ddnet] Pull request opened: #5186 Disable IPO by default since it slows down linking a lot, Use better available linkers (mold > lld > gold > ld)

I'll still build the official builds with IPO=ON. CC @Kaffeine

Checklist

[X] 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 mem...

23:37

[ddnet-scripts:master] 1 new commit

cfd5ace IPO ON - def-

Exported 382 message(s)