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 2022-07-15 00:00:00Z and 2022-07-16 00:00:00Z

little idea I had, still needs a bit of work but is it something we'd wanna add?

Image attachment

Ravie

little idea I had, still needs a bit of work but is it something we'd wanna add?

would be nice for midground scaling

bluesky

@ The Windows Users, what IDE are you all using for C/C++ ? (edited)

eclipse best

1

Ravie

little idea I had, still needs a bit of work but is it something we'd wanna add?

yes please

i use visual studio / microsoft compiler on windows. i use visual studio just for building. vscode for coding. with the c++ extensions pack by microsoft. which is actually only available in the microsoft branded version of vs code i think. i also tried to build ddnet with mingw in msys or in wsl or i tried the cmake gui, but didn't make the cmake gui work. but with visual studio it's just one click. i don't care about the bloat xd

02:06

aaand: that said, i actually am eager to use mingw / vim / neovim / any open source setup instead of vscode. and also not use electron.

02:10

the one thing that i not get done is accessing files in a big projcet with vim. in ides you simply have a file tree. or ctrl + p. but in vim i open files from terminal. i know there is nerdtree, which maybe could be useful. or fzf. but i kind of feel lost with only vim in a big code base. but i feel lost in a big ide with ten menus in the top bar with 20 undermenus xd

cauldron

i use visual studio / microsoft compiler on windows. i use visual studio just for building. vscode for coding. with the c++ extensions pack by microsoft. which is actually only available in the microsoft branded version of vs code i think. i also tried to build ddnet with mingw in msys or in wsl or i tried the cmake gui, but didn't make the cmake gui work. but with visual studio it's just one click. i don't care about the bloat xd

thats alot of information

02:15

thanks for the answer

02:16

i heard of all that stuff but never used it

02:17

i guess i will stick with VS for now since i mainly use windows. I will try the other Editors at some point (edited)

02:18

=))))

02:18

VS is just a great all-in-one solution

i agree

Is this how ENTTYPE_LASER works? Or is every laser just the same and is server-sided only when it comes to difference between lasers?

Image attachment

Voxel

Is this how ENTTYPE_LASER works? Or is every laser just the same and is server-sided only when it comes to difference between lasers?

Every laser is same, but on server side it have different functions

Is there a way for the client to communicate with the server to determine which laser is which?

Voxel

Is there a way for the client to communicate with the server to determine which laser is which?

you mean vice versa? for the purposes of differentiating them i imagine

lynn

you mean vice versa? for the purposes of differentiating them i imagine

Yea

i suppose it could be done in tandem with different settings for the colors. could create a github issue for it

05:34

i assume you mean the desire to change the color per type

Yeah. The plan in my head is there for there to be 3 or 4 different types: Rifle, Shotgun and Game/Freeze

Voxel

Is this how ENTTYPE_LASER works? Or is every laser just the same and is server-sided only when it comes to difference between lasers?

vanilla compat stuff

cauldron

i use visual studio / microsoft compiler on windows. i use visual studio just for building. vscode for coding. with the c++ extensions pack by microsoft. which is actually only available in the microsoft branded version of vs code i think. i also tried to build ddnet with mingw in msys or in wsl or i tried the cmake gui, but didn't make the cmake gui work. but with visual studio it's just one click. i don't care about the bloat xd

You can get the extension pack in codium by using the microsoft store btw

cauldron

the one thing that i not get done is accessing files in a big projcet with vim. in ides you simply have a file tree. or ctrl + p. but in vim i open files from terminal. i know there is nerdtree, which maybe could be useful. or fzf. but i kind of feel lost with only vim in a big code base. but i feel lost in a big ide with ten menus in the top bar with 20 undermenus xd

There is also the command-t plugin for vim, which is basically ctrl-p from other editors

Voxel

Is there a way for the client to communicate with the server to determine which laser is which?

You'd have to come up with that one. One could maybe try to sneak a entity subtype or atleast a laser subtype somewhere like we snuck the owner and flags into projectiles

Voxel

Is there a way for the client to communicate with the server to determine which laser is which?

yes, thats what EntityEx is for

Learath2

You'd have to come up with that one. One could maybe try to sneak a entity subtype or atleast a laser subtype somewhere like we snuck the owner and flags into projectiles

no. and please not more of this. never again.

Ah we do have a proper way now, forgot we added that

Can one fix the projectiles btw

Wym by fix?

Its sad that I can not use them properly due to the fact that high velocity makes the client think its a packed one and therefore its completely broken

Hm, I thought the extended flag was high enough that you'd never really want a projectile that fast

fokkonaut

Its sad that I can not use them properly due to the fact that high velocity makes the client think its a packed one and therefore its completely broken

wym with packed (edited)

I thought so too, but it happes due to the way I use projectiles to fake them, with faked velocities in order to have them stable and be shown good on the client. For prediction.

bluesky

wym with packed (edited)

There is extra info packed into the normal data fields, which the client will use if a specific flag is activated

okay what would be an example for the extra data

10:27

well the code overall is pretty divers and a few comments that say "do this better and nicer" and "todo" and such so yeah there is still alot to optimise

10:28

but i dont wanna touch those until i am more familiar with the whole codebase (edited)

10:29

i started 2 weeks ago or so

10:30

oh well no, not even 2 weeks XD

10:31

this channel is a good place to learn, i love it

More like this chanbel is the only place to learn justatest

justatest

Ryozuki

More like this chanbel is the only place to learn justatest

justatest

there are also indian tutorials on YT

About ddnet?

probably

Ryozuki

More like this chanbel is the only place to learn justatest

justatest

i dont talk about ddrace specifically. i mean dev in general (edited)

cauldron

the one thing that i not get done is accessing files in a big projcet with vim. in ides you simply have a file tree. or ctrl + p. but in vim i open files from terminal. i know there is nerdtree, which maybe could be useful. or fzf. but i kind of feel lost with only vim in a big code base. but i feel lost in a big ide with ten menus in the top bar with 20 undermenus xd

Vim has a ctrl p plugin to search files

10:35

Using fzf

10:36

My neovim uses fzf + fd (made in rust, faster)

10:37

And it previews files, highlighted with bat (made in rust too, like cat)

10:37

Its probs way faster

@Ryozuki are you coding on linux?

Off

10:38

Ofc

yo guys, did u remove cl_show_quads on last nightly?

Im not a masochist

Ryozuki

Im not a masochist

well true except if you mainly develop .net apps

szaro

yo guys, did u remove cl_show_quads on last nightly?

They didnt follow their own rules about not renaming options for no reason, because players would be confused, so its cl_showquads now.

10:40

DDNet people do many things that they wouldnt allow for others, they mostly just do things out of the moment :)

Yes yes go take ur childish grudge elsewhere

Ok boomer

okboomer

szaro

yo guys, did u remove cl_show_quads on last nightly?

https://github.com/ddnet/ddnet/pull/5511

Rename cl_show_quads -> cl_showquads by def- · Pull Request #5511 ·...

To reenable quads for everyone who accidentally disabled it or has the bind to accidentally disable them Checklist Tested the change ingame Provided screenshots if it is a visual change Tested...

bluesky

well true except if you mainly develop .net apps

Mono isn't too horrible nowadays

Learath2

Mono isn't too horrible nowadays

mono is great

But yes, if I had to develop .net apps all the time, I would also be on windows

happy

1

fokkonaut

They didnt follow their own rules about not renaming options for no reason, because players would be confused, so its cl_showquads now.

It's not for no reason, it was to reset it. Ideally we'd have another way to do that, but we don't

why was the bind not removed then

11:08

xd

fokkonaut

why was the bind not removed then

i mean people still use it, it was just a reset for everyone

11:10

"To reenable quads for everyone who accidentally disabled it or has the bind to accidentally disable them" def-

yea, "or has the bind to accidentally disable them"

i dont know people with a bind for that

Wasnt there a standard bind for it?

11:11

Well idk, for me it seems useless to reset it

how can i unbind ctrl+mouse2

11:26

in editor it makes me swap from game layer to some design layer always and its super annoying

fokkonaut

Wasnt there a standard bind for it?

i think pagedown maybe

szaro

yo guys, did u remove cl_show_quads on last nightly?

renamed to cl_showquads. Why do you use it?

fokkonaut

DDNet people do many things that they wouldnt allow for others, they mostly just do things out of the moment :)

How about asking for the reason instead of assuming the worst?

12:01

I keep getting bug reports where people accidentally turned off quads and didn‘t realize what happened. This fixes it for them while still allowing people who have a use case for quads off to use it

But how can this fix it long term? The same thing will happen again

12:08

especially if we dont remove the default bind (as i said idk if it stil lexists)

deen

renamed to cl_showquads. Why do you use it?

it's on pagedown by default and i sometimes use it when i want to play with textures but dont want the background to distract me as much

i think there are not that less people using it. But since u can easily adapt your entities and background to make any map look like u want it may exstinct

fokkonaut

But how can this fix it long term? The same thing will happen again

No, we don't have the bind anymore by default

12:29

The problem was that before we had a default bind for it, so people would just randomly click it. This is the only way to remove a bind from everyone's config

szaro

it's on pagedown by default and i sometimes use it when i want to play with textures but dont want the background to distract me as much

Fair enough, but some maps use quads for game relevant stuff

I have my own quad bind and one for high details, its really good to be able to have "different themes" depending on which map and design, cause some maps cant be played in entities, when there are just insanely many tiles, or the old hookthrough i specific places, then i prefer to play without and turn off quads or so, depending on how ugly or unplayable the real design is

I really want to do something about default configs. Maybe we could only save ones that are different from the default instead of all settings? People getting stuck with old settings is so annoying

isnt that the case already? at least i somehow thought so

No, the client will save the entire set of settings, so you'll get stuck with old defaults

Ah, well idk but this wouldnt have fixed this issue anyways

Maybe something over info.ddnet.tw?

Yea

Something like a config update pushed over the network. Each update could have a serial and the config serial would be bumped after each update. That way after a config update gets pushed out people can still change it back and it wouldnt get overridden

then we could also rename configs without problems, when the new info gets fetched the client could set the new config to the old value

12:39

(optional, of course not in cases like this, but in general)

Yep, that could also be achieved, some sort of OTA update capability and not saving defaults might be nice

haha deen i remember when Cloudly died in Lava at Lockdown tournament and went really mad because it was invisible without quads

Learath2

Yep, that could also be achieved, some sort of OTA update capability and not saving defaults might be nice

agree

but yeah, some old maps have looping kill tiles for example, u wont see them. Wont happen in new maps because people will be told to put some marking below the quad

[quakenet] Konsti

haha deen i remember when Cloudly died in Lava at Lockdown tournament and went really mad because it was invisible without quads

I have a speedup made of moving quads on my server, which is pointing downwards, and for people without quads on I placed a tile layer behind the background because that is also not shown when quads are off, so that they can see tile layer speedups on that spot

12:44

testers should look for stuff like that and especially point that out and force stuff like this

yes they do. Lockdown was released due some mafious structures :justatest:

ah ye

12:45

do they do nowadays?

of course. We already had the case. its also possible to animate tiles without quads

i know, but not moving, just color

How is intels website so bad with some of the best engineers on earth??

Ryozuki

Vim has a ctrl p plugin to search files

i know i also have fd and fzf. but i need a visual filetree maybe to see the project. im not good at having all my vim plugins in place. i actually try to avoid them like oh-my-zsh or vim plug. maybe i can make it work with a single neovim config. i lately put visual representation for tabs and spaces and always have spaces. i like this and its only a setting. maybe there is a way to make vim/nvim show the tree of the root folder of the git repo you're in right now. without plugin. arent there panes? so have a 10 or 20 % pane on the left with nvim showing the filetree without distracting by accidently entering into this pane. i like this idea i try it

15:08

i mean why plugins, mostly you can get similar functionality without by configuring, at least try

15:10

i want to.try

U dont need file trees honestly

15:11

I also thought like that

15:12

Before becoming :based:

15:12

Stupid mobile

15:12

PES2_BasedGe

ok

Learath2

How is intels website so bad with some of the best engineers on earth??

all the budget goes into chip making?

It's absolutely bizarre how shit this website is. How hard is it to code a login page?

i only use https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html on their website

Their licensing system for the FPGA software is so horrible. This is the reason I hate proprietary software. They can't even do it properly

what you doing with FPGA and VHDL?

Konsti: pink rat

Chairn

what you doing with FPGA and VHDL?

uni course on computer architecture has a bit of lab work involving fpgas

15:21

not absolutely necessary but I've always been interested

15:22

I already hate it thanks to intel

i've never used any intel stuff for fpga

15:26

or did they completely incorporated altera stuff into intel website?

I think intel did acquire all of altera and merge it into their own site

15:27

it's now intel quartus and intel questa

I spent an hour trying to simulate a design to no avail, until I found some random stackoverflow post where a guy suggested adding -voptargs="+acc" without any explanation, added it works perfectly

16:21

Isn't proprietary software like this supposed to be documented oh so better than open source hypervomit

hypervomit

software?

In general, i use -novopt

16:24

slightly slower, but should be more robust

I was using that too, but -novopt option is now deprecated and will be removed in future releases.

oh

funnily enough -novopt is actually the default still in quartus, so if you just open a waveform and click simulate it just won't work

16:25

and you are left wondering for an entire hour what went wrong

16:25

How do 2 pieces of software shipped by the same company, possibly charging thousands for a proper license not work together perfectly

you need the extra support license for only 10k$

16:27

that's why companies usually lag a few years behind on production software

If I was paying 10k$ and I got errors while simulating a full adder I'd be furious

during my phd, we were using version from 2010

16:27

worked fine

But finally, my 5 bit adder can be simulated

Image attachment

16:32

Ngl I'm a little disappointed, when I heard the class and the material were based on a 7gb proprietary editor I was a little irked but I was happy atleast it'd all justwork(tm) and I could focus on doing things rather than setting up things

16:33

I spent an hour trying to get it to install, another hour trying to license it and another hour digging in awful documentation trying to find what they replaced that one deprecated option with only to find it on stackoverflow

Learath2

But finally, my 5 bit adder can be simulated

next step is now sequential logic

16:37

then you can reimplement ddnet in vhdl

16:37

with VGA output

I was really excited for it when I started yesterday but now I'm just a little exhausted

16:38

I was going to make both a CLA and a sequential adder

how come you have classes in summer?

16:38

sequential adder is kinda useless

Chairn

how come you have classes in summer?

I don't, I'm the laziest person on the face of earth, so I didn't study when I was supposed to

kek

Now I need to learn all the material everyone else learned over 4 months, in 1 month

16:39

The only saving grace is that I already know a lot of the basics unlike the people taking the class

16:40

I can also skip the part they learn some motorola 68k assembly, I'm already familiar with the ISA

did you learn about propagation delay yet? (edited)

Yep

16:42

Not as part of this class though, I learned about some of this content 5 years back when I was building an 8bit cpu on a breadboard

16:43

Obviously inspired by that video series on youtube by ben eater

16:43

https://youtube.com/playlist?list=PLowKtXNTBypGqImE405J2565dvjafglHU quite fun as a watch and the guy is pretty great at explaining stuff

Building an 8-bit breadboard computer!

lol, crazy guy

it's actually surprisingly fun to make and you learn a lot along the wya

16:46

way*

16:48

Actually, I think the first time I encountered propagation delay was probably in minecraft. The lack of propagation delay there made some interesting circuits possible

okay, weird stuff: https://github.com/Chairn/ddnet/commit/5adc3d51d9654ce03eb5b2f2af3e685fd2ac37d5 Why does this crash puzzles me...

16:51

both server and client

want me to check it out or did you already figure it out?

i didn't figure it out

16:53

i guess that somehow, a size is implied somewhere

16:54

this is client backtrace:

    #0 0x7b7e05 in CLayers::InitTilemapSkip() /nobackup/opt/ddnet/clangbuild/../src/game/layers.cpp:191:11
    #1 0x7b98f4 in CLayers::InitBackground(IMap*) /nobackup/opt/ddnet/clangbuild/../src/game/layers.cpp:167:2
    #2 0x1361a8f in CMenuBackground::LoadMenuBackground(bool, bool) /nobackup/opt/ddnet/clangbuild/../src/game/client/components/menu_background.cpp:267:15
    #3 0x135e576 in CMenuBackground::OnInit() /nobackup/opt/ddnet/clangbuild/../src/game/client/components/menu_background.cpp:50:3
    #4 0x17daa6e in CGameClient::OnInit() /nobackup/opt/ddnet/clangbuild/../src/game/client/gameclient.cpp:265:15
    #5 0xcee84a in CClient::Run() /nobackup/opt/ddnet/clangbuild/../src/engine/client/client.cpp:3143:16

https://github.com/ddnet/ddnet/issues/1 https://github.com/ddnet/ddnet/issues/2 https://github.com/ddnet/ddnet/issues/3 https://github.com/ddnet/ddnet/issues/4 https://github.com/ddnet/ddnet/issues/5

..., i thought this linking issue thing was removed?

16:55

and server backtrace:

    .#0 0x79ae55 in CLayers::InitTilemapSkip() /nobackup/opt/ddnet/clangbuild/../src/game/layers.cpp:191:11
    .#1 0x799cb6 in CLayers::Init(IKernel*) /nobackup/opt/ddnet/clangbuild/../src/game/layers.cpp:115:2
    .#2 0xb37697 in CGameContext::OnInit() /nobackup/opt/ddnet/clangbuild/../src/game/server/gamecontext.cpp:3210:11
    .#3 0x86b7fe in CServer::Run() /nobackup/opt/ddnet/clangbuild/../src/engine/server/server.cpp:2631:16

https://github.com/ddnet/ddnet/issues/1 https://github.com/ddnet/ddnet/issues/2 https://github.com/ddnet/ddnet/issues/3

lol the first github issue has tons of unintentional references (edited)

meh, I thought that was fixed :/

16:57

@Chairn those look like parts of the map format btw, I can see that breaking things actually

16:58

a vector4_base<int> probably wouldn't break

it was int before, but colors are actually unsigned chars so i thought reducing memory footprint

16:59

but yeah you're right, it's probably due to map files hence both crash

Chairn

it was int before, but colors are actually unsigned chars so i thought reducing memory footprint

We probably use that CColor directly as is inside the map format :/

const char * const CSkins::ms_apColorComponents[NUM_COLOR_COMPONENTS] = {"hue", "sat", "lgt", "alp"};

18:06

why would you use the modifier two times in C++ code? (general question) (edited)

first const if for the pointed data

18:06

second const if for the pointer itself

okay

18:07

thx

18:09

you can override a const pointer, so why would you still use the modifier const infront of it ? (edited)

it is a member variable which gets set here

example:

const CSkin::SSkinTextures *pSkinTextures = pInfo->m_CustomColoredSkin ? &pInfo->m_ColorableRenderSkin : &pInfo->m_OriginalRenderSkin;

render.cpp

const char *data --> data is immutable but may point to wherever you want

18:12

const char * const data --> data is immutable and the pointer cannot point to anywhere else

1

thanks^_^

Chairn

const char *data --> data is immutable but may point to wherever you want

sry i still got a question

Chairn

const char *data --> data is immutable but may point to wherever you want

in what sense is a non const * less "immutable" than a const *

in the first case, you can assign to whatever you want: data = &otherdata;

18:15

in the second, you cant

18:16

or i don't get your question?

like even another type

Chairn

or i don't get your question?

nah, i am just bad at explaining sry const CSkin::SSkinTextures *pSkinTextures CSkin::SSkinTextures *pSkinTextures (edited)

18:16

i can do the same thing with both

no, first case you cannot modify the pointed data

do you mean i can assign whatever i want to the latter, even another type?

Chairn

no, first case you cannot modify the pointed data

ay wait, discord removed the * for the pointer OOOOOOOOOOOOOF

for instance, you cant do:

c++
const char* pData = &otherdata;
str_format(pData, "%s", "abcdef");

18:18

because pData points to const data

18:18

but you can do:

c++
char aBuf[] = { 'a', 'b', 'c' };
const char* p1 = aBuf;
p1[0] = 'z'; // <-- error
char* p2 = aBuf;
p2[0] = 'z'; // <-- no error

(edited)

18:20

however you are right too i think

18:20

if u do it with a struct you can copy it into a non const ptr

18:23

example

c++

struct a {
 int* p;
};

struct a tmp;
const a *pPtr = &tmp;
int* pPtrModify = pPtr->p;

*pPtrModify = 1;

will probably work

18:23

bcs you aren't modifying the struct

18:23

just what it points to

OK

18:26

Looks all weird to me

18:26

but i kinda get it

Jupstar ✪

c++
char aBuf[] = { 'a', 'b', 'c' };
const char* p1 = aBuf;
p1[0] = 'z'; // <-- error
char* p2 = aBuf;
p2[0] = 'z'; // <-- no error

(edited)

this makes sense to me

18:27

OKAY I GET IT NOW

Something that helped me understand consts: const X is actually syntactic sugar for X const, meaning const is usually on the right of what it qualifies. and const char * const is the same as char const * const, const std::string& is the same as std::string const&…

I think now understand it: You can change where it points to but you cant modifie individual member values Am i on the right path? @Jupstar ✪ @Chairn (edited)

that way you can better see whether it's the char part or the * part which is const

k2d222

Something that helped me understand consts: const X is actually syntactic sugar for X const, meaning const is usually on the right of what it qualifies. and const char * const is the same as char const * const, const std::string& is the same as std::string const&…

you mean my last question i guess. i understood that already. Thanks tho! happy

happy

(edited)

bluesky

I think now understand it: You can change where it points to but you cant modifie individual member values Am i on the right path? @Jupstar ✪ @Chairn (edited)

yes

brownbear

1

Learath2

I really want to do something about default configs. Maybe we could only save ones that are different from the default instead of all settings? People getting stuck with old settings is so annoying

we only save those that are different from default

18:43

but that's not a good solution either

18:43

that makes changing config defaults bad if we don't want to change it for existing configs

18:43

we should always save all configs, and make explicit migrations

heinrich5991

we only save those that are different from default

Are you sure? We did discuss it but I don’t remember it getting implemented

yes, sure

18:44

we have a discussion about it when we change config defaults

18:44

because suddenly that's something to discuss, because it changes existing players' configs

heinrich5991

that makes changing config defaults bad if we don't want to change it for existing configs

Hm, I can’t imagine not wanting to change configs for players that haven’t touched them

18:45

We do want everyone on what we think is the sane default unless they explicitly opt out, no?

it's unclear if they have touched it or not

18:45

they just changed it to default again or not

18:45

think e.g. default fullscreen setting

18:45

I'm sure most people set it to what they prefer

18:45

if we change it, it'll change for everyone who was on the old default

18:46

that's really annoying (edited)

That’s a small issue, no? They can just change it back. It’s desirable that everyone who doesn’t care ends up with good defaults

heinrich5991

we should always save all configs, and make explicit migrations

you achieve that with this ^

18:47

but without the downsides

18:47

check the discussions where we change defaults

Learath2

That’s a small issue, no? They can just change it back. It’s desirable that everyone who doesn’t care ends up with good defaults

even tho i agree, the defaults on many linux apps suck xDD

heinrich5991

you achieve that with this ^

Does that really fix it? If we change a default and make a migration for it, the people who would be annoyed by the default change will get annoyed

it somewhat fixes it because we can then explicitly decide whether to change existing players config

18:48

for something like fullscreen, that'd probably be a "no, don't change"

Hm, but that doesn’t help with the people that don’t care getting good defaults. We make a migration it’ll migrate everyone

18:53

Say we want to make vulkan default, we push a migration for it, konsti gets annoyed because he really wants opengl 1.4 for whatever reason

I guess with that approach, we could even let people opt out of config migrations?

fokkonaut

yes, thats what EntityEx is for

that's one step closer to what I'm searching for! Now all I need to figure out is how Rifles and Shotguns connect into the equation!

wym?

My plan is to have make something like this. Doors and Freeze Lasers will also have their own color(s), but my main focus is making it so that the ability for different colored lasers is possible.

Image attachment

nice

20:21

i was also thonking of different laser colors in fng for different teams

cauldron

i was also thonking of different laser colors in fng for different teams

that would be fun, but i would have to figure how to do use teams in this

its not gonna happen.i think

20:26

its server feature i believe

i mean this exists

Image attachment

ill check, or maybe someone here knows

Maybe I'm thinking about this whole thing wrong. What if instead of focusing on specifics, I could just create 4 abstract "modes" with laser colors. Given the names will be the same, it would save the hassle of trying to loop around what is what. And instead call for a mode when it's needed by one laser.

you mean make all laser brown when shotgun is selected? then other people with real laser would shoot brown laser too

Voxel

My plan is to have make something like this. Doors and Freeze Lasers will also have their own color(s), but my main focus is making it so that the ability for different colored lasers is possible.

nice

cauldron

you mean make all laser brown when shotgun is selected? then other people with real laser would shoot brown laser too

does the server not recognize other people's weapons?

Voxel

Maybe I'm thinking about this whole thing wrong. What if instead of focusing on specifics, I could just create 4 abstract "modes" with laser colors. Given the names will be the same, it would save the hassle of trying to loop around what is what. And instead call for a mode when it's needed by one laser.

I think best is if we do this in a kinda mod friendly way. Either, we choose the color based on the entity type (shotgun, rifle, door, freeze wall), or a separated thing to tell the color.

20:33

You dont have to struggle with weapon holding or whatever

I was thinking the first thing, but if the second is more reccomended, then i could go with that (edited)

It'll probably be the first thing

Alright

Idk the best way to do it, but we should probably try to make it in a way that provides some prediction information too so it's not completely wasting bandwidth just for visual stuff

yep

not really a waste if it creates different visuals, but I agree that having prediction information is nice, too

Huh, EntityEx is surprisingly rigid wtf

yep

How did we get it so wrong with such a recent addition?

thats what im wondering every time, these kinda things just get merged

20:37

and i always get told to check explicitly and to talk before and so on

fokkonaut

thats what im wondering every time, these kinda things just get merged

I remember we talked hours about EntityEx

but not enough :d

20:37

and: we talked about it when it was too late already (?)

I'm sure we must have been thinking something when we accepted such a generic name for something that looks so rigid

Also the way it works is not really nice.

20:39

It creates a separate snap object for each entity

after all you can.do anything client side (edited)

#4236

https://github.com/ddnet/ddnet/pull/4236

Reduce cpu usage from large numbers of entities + EntityEx by trml ...

An attempt to fix #4233. The problem was (as guessed) that the contains a very large number of laser doors, which each has an EntityEx. Calling SnapFindItem (which does a linear search) for each on...

EntityEx seems to be about static entities in the map(?)

@Learath2 doesnt change anything about my sentence

#4176 actually

https://github.com/ddnet/ddnet/pull/4176

Entity switch prediction by trml · Pull Request #4176 · ddnet/ddnet

Adds the prediction for pickups and projectiles (based on #4108) For blinking entities (deactivated by switch) I also used the extended info to make the blinking effect happen client-side. So inste...

fokkonaut

It creates a separate snap object for each entity

this is desired imo

Well... Its okay in general, if we had extended snapshots

extended = larger?

yes

@heinrich5991 hm, that does seem to be the case. Should have been names StaticEntityEx

you can rename it @Learath2

heinrich5991

extended = larger?

I was thinking a SNAP_MORE where we can put the overflow instead of dropping it

20:42

We can't have variable sized stuff in the snap, can we?

[ddnet/ddnet] Issue opened: #5632 Get rid of `validate_size=False` in datasrc/network.py

I haven't investigated, but we recently merged some changes that allowed incomplete snap objects by filling some defaults. This mechanism should be made explicit: Mark the fields that are optional and that are automatically filled.

@Learath2 hm, I'm not sure if our snap diffing works for an object that changes size

20:44

but I think it could be made to work

Now that I think about it, there is no harm in introducing new extended types, right? we are not constrained on space that way

we are not constrained in any way (edited)

We don't need this to be very generic, we can just have a DDNetLaser like we have a DDNetProjectile

heinrich5991

we are not constrained in any way (edited)

We are a little tight on snap space, some of the larger maps were having issues before we started clipping more aggressively because of the sheer amount of EntityEx's

Learath2

We don't need this to be very generic, we can just have a DDNetLaser like we have a DDNetProjectile

I think all of this would make it too complicated no?

On the contrary, I thought it'd be simpler than a more generic DynEntityEx for entities that are spawned at runtime

hm

20:49

i have the feeling we can soon simply create a new Snap function called SnapDDNet()

We can have a NetInt or NetEnum for laser type so we can get prediction information, and we can have a NetInt for color

or

20:50

we rename the old one to LegacySnap

and thanks to recent work our clients can handle netobjects getting larger

for older clients + 0.7

Trying to split that up leads to the same issues I've had while trying to keep the 0.7 compatibility stuff split. As you go down the callstack you lose more and more information about specifics

20:52

s/down/up/

so i tried doing this for the menu just to see if i could create a mockup for preview rendering turns out this softlocked my computer into fronting the frozen client and i had to restart. was this not a valid shortcut?

Image attachment

Image attachment

22:08

inb4 im called yanderedev

IsRifle = false

22:08

is that a single equals?

oh fuck

22:09

fixed

it looks like you might want to use a different font

22:09

it looks hard to distinguish in your font ^^

i'm using Fira Code

Your font is too fancy, it replaces -> with an actual arrow, == with a long =. That looks slightly confusing

I mean the arrow isn't that bad, but the equals apparently leads to problems

do you have font reccomendations?

no, I mostly use defaults

22:12

I think my terminal uses dejavu sans mono or so

I just disabled ligatures. It's a bit better now.

Image attachment

weird colors

YES

Image attachment

nice

22:24

little tip, you don't need == true or == false on bools

Ohh. That's useful.

Is there anywhere a concrete list of symbols you can write in chat?

is there any symbol you can't write in chat?

I'm just wondering on what types of emotes you can write or anything for that matter

ah

22:42

idk, you'd probably have to look at the font for that

what's the font?

Voxel

I just disabled ligatures. It's a bit better now.

much better

estguy

what's the font?

dejavu sans I think

DejaVuSans.ttf

22:47

there's also Icons.otf, but i don't know what it contains

there's also some fallback font for asian languages

how to I make the two weapons stack and not overlap?

Image attachment

Voxel

how to I make the two weapons stack and not overlap?

do another HSplitTop I think

Image attachment

23:01

Nowhere near close to the final product, but this is a solid head start for what I want the idea to head to.

@bluesky progress

which code do you prefer ?

c++
    new(pDemoHeader) std::remove_pointer<decltype(pDemoHeader)>::type;
    new(pTimelineMarkers) CTimelineMarkers;

23:22

first line ensure type is safe but is more verbose

23:23

second line makes no guarentee on the type if pTimelineMarkers type is changed, new will override neighboring memory or not initialize enough (with the wrong parameters)

can't you do decltype(*pDemoHeader) in the first case?

no,

23:28

https://stackoverflow.com/questions/25309356/using-new-with-decltype and https://stackoverflow.com/questions/17241614/what-expressions-yield-a-reference-type-when-decltype-is-applied-to-them/17242295#17242295 for why

23:29

error: new cannot be applied to a reference type

23:30

roughly, decltype(*pDemoHeader) returns a reference to a class (here CDemoHeader&) which cannot be passed to new...

ah, unlucky

Voxel

so i tried doing this for the menu just to see if i could create a mockup for preview rendering turns out this softlocked my computer into fronting the frozen client and i had to restart. was this not a valid shortcut?

The hard quit shortcut ctrl+shift+q should always work iirc

didn't that shortcut get removed because of trolling?

Hm, if it did then you are stuck, which is annoying

23:38

Maybe we should add it back with a #ifdef DEBUG it's so useful in debug builds

yeah closing the game is annoying when you're in fullscreen and get an exception

looks like it got removed

23:40

it has no effect on latest master

Should definitely get it back in for debug builds, it saved my ass so many times while debugging SDL issues on macOS

Chairn

which code do you prefer ?

c++
    new(pDemoHeader) std::remove_pointer<decltype(pDemoHeader)>::type;
    new(pTimelineMarkers) CTimelineMarkers;

first form also works for array which is nice

Literally the only way out is that hard bind

alt f4 is always a thing

Chairn

which code do you prefer ?

c++
    new(pDemoHeader) std::remove_pointer<decltype(pDemoHeader)>::type;
    new(pTimelineMarkers) CTimelineMarkers;

I like the first one, but the second one is definitely prettier

df82d24ff (deen/pr-ctrl-shift-q) Remove panic quit button (1 year, 11 months ago)

Ravie

alt f4 is always a thing

when messing with things like SDL you can cause that event to get eaten aswell accidentally

https://github.com/ddnet/ddnet/pull/2650

23:42

ahahahha, you accepted the PR @Learath2

heh, I guess the alt+f4 argument convinced me back in the day

why would it not work anyway?

23:43

if main loop is running, it should work?

well on mac once I was messing around with raw input, I ended up in a state where I was stealing all the inputs

ddnet as keylogger, nice one

23:45

did you manage to steal your own password

?

23:45

It was more to try to get the mouse to capture, but even with raw input it didn't look very good so I gave up on that

23:46

Another weird issue on mac again is real fullscreen, if you launch the game in real fullscreen it'll launch just a blackscreen with a non-running appdelegate, so all your cmd+q are going nowhere

23:47

But yeah, outside of mac weirdness I guess the WMs kill shortcut should just work

23:48

on linux even if that doesn't work you can drop down to a tty, on windows I think ctrl+alt+del doesn't even go through the WM, the kernel is catching it iirc

Exported 395 message(s)