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 2022-07-16 00:00:00Z and 2022-07-17 00:00:00Z

Voxel

Click to see attachment 🖼️

wowow looks so cool cant wait

Voxel

Click to see attachment 🖼️

nice

Voxel

@bluesky progress

hm?

i was linking you to the photo

[ddnet-maps:master] 1 new commit

8939e7d M Just Pseudo Fly - ddnet-maps

Chairn

which code do you prefer ?

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

I prefer std::make_unique

I'm getting lots of these in the console

Image attachment

09:21

I can't find any references to the numbers in steam documentation idk what they mean

stem

deen

I prefer std::make_unique

This is not the same, im using placement new to initialize objects rather than mem_zero. Im not allocating memory

why even use new

10:33

*pPtr = CClass()

you can do that?

10:34

i never tried

does new(buffer) CClass even call the destructor?

10:35

sounds rather dangerous to experiment with such things if you not sure what u doing

placement new just call the constructor on the given address

so the old object stays active?

10:35

or is not properly destructed

old object is overridden

*

so not properly destructed which was already the case before with mem_zero

so its no improvement over mem_zero

10:36

u just make it just as bad xD

well, they don't have destructor anyway, but i can still call it before if you want

10:39

looks like i can't call constructor:

 error: invalid use of ‘CNetConsole::CNetConsole’
   11 |         this->CNetConsole()

Jupstar ✪

*pPtr = CClass()

this create new object on the stack rather than reinitializing old object, so not the same behavior

10:41

and the object is definitely destructed at the end of the function but it should live longer (edited)

Use smart pointers everywhere

10:41

PES4_MadGe

Chairn

this create new object on the stack rather than reinitializing old object, so not the same behavior

?

10:42

where is the stack variable?

CClass() is stack variable

but there is an asignment

10:43

=

Whats this new() thing as the type

oh

10:43

stupid me

also its not a variable, its a pure r value

lel, assignment much better

Ah places an object into allocated memory

Ryozuki

Whats this new() thing as the type

i don't understand your question

The placement new thing

10:44

Never heard of it till today xd

@Jupstar ✪ what would you do for array of objects?

10:45

placement new can init array but assignment requires loop

it will also loop internally

10:46

bcs all objects will call constructor

i know, im just lazy to type the loop

dunno use foreach or smth

[ddnet/ddnet] Pull request opened: #5633 Adapt upstreams CButtonContainer

Introduce CButtonContainer which is used for fading text in teeworlds. There is no fading text in ddnet yet. This change allows introducing such fading buttons but the main motivation for this change is easier copy & pasting code from ddnet to teeworlds code base.

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 (especially base/) or ad...

[ddnet/ddnet] Pull request opened: #5634 Fix clang tidy accessing data pointer

ddnet/src/engine/client/favorites.cpp:229:23: error: 'data' should be used for accessing the data pointer instead of taking the address of the 0-th element [readability-container-data-pointer,-warnings-as-errors] int Index = pEntry - &m_aEntries[0]; ^~~~~~ (m_aEntries).data()

Checklist

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

Voxel

Click to see attachment 🖼️

so you are done with that? or is it just the menu yet, where you can define every color in its own preview function?

well, clang does not know loop and gcc kinda store variable per variable:

16:42

https://godbolt.org/z/6d17Teq6W (edited)

I do not update the list of servers (there is an eternal update), but only on windows 8.1, before that was windows 7 and everything worked fine, and before windows 7 was windows 8.1 and it also does not update, I think the problem is somehow connected with the system

17:03

[ddnet-web:master] 1 new commit

68261af Remove HappyFray from mods as he resigned - def-

tear

3

"xd sure chillerbot.png is lyfe" lol

@Tater damn before I went away a week ago I opened your shadertoy page. now that I'm back that was the first page I have open and I'm absolutely amazed :D

cauldron

so you are done with that? or is it just the menu yet, where you can define every color in its own preview function?

its just the menu

i hate the fact that std::size cannot be used in a constexpr when the array is stored in a struct in a vector...

hm, why?

Chairn

i hate the fact that std::size cannot be used in a constexpr when the array is stored in a struct in a vector...

show code or did not happen

c++
constexpr int ArraySize = std::size(Container.m_vQuads[0].m_aVertices); // operator[] is not constexpr so compilation fails
constexpr int ArraySize = sizeof(Container.m_vQuads[0].m_aVertices)/sizeof(Container.m_vQuads[0].m_aVertices[0]); // ok

Ah, I see what you mean

so its not about std::size or structs, but only about accessing an array

i tried everything

20:18

you cant use :: on non static member

constexpr reference operator[]( size_type pos ); (since C++20)

so i had to resort to the good old sizeof

c++20 bro

i wish we used c++ 20, i need auto generated operator==

all versions before c++20 just suck no joke xD

20:19

and in 1 year i say the same with c++23

i wonder what c++ will look like in 20 years

in 20years KIs are coding what we want

20:20

we just type in the idea

finally they will have implemented everything that people needed since 2000 :D (edited)

20:20

ki?

AI

what kind of keyboard do you have?

20:21

a and k are opposite side

its the german shortcut

ah

i am just used to call it KI xD

artificial = kartificial?

künstliche

in french its even worse, we use IA instead of AI

20:22

so i always mix them up

@Chairn constexpr auto k = std::tuple_size<decltype(decltype(v_vec)::value_type::m_arr)>::value; maybe?

Learath2

@Chairn constexpr auto k = std::tuple_size<decltype(decltype(v_vec)::value_type::m_arr)>::value; maybe?

yeah i saw that, but i thought that sizeof should be enough

20:23

uh, the stackoverflow i saw had only one decltype

use sizeof, put a comment TODO: c++20 u nobos

20:23

and its fine

Jupstar ✪

constexpr reference operator[]( size_type pos ); (since C++20)

I'm not sure if this is even fine in C++20 actually

20:24

Isn't that constexpr operator only available if the vector itself is constexpr?

possible, but he uses it in a constexpr, so i assume it is?

20:25

or why is his function constexpr in first place

Jupstar ✪

or why is his function constexpr in first place

The size of the array inside the struct inside the vector is constant at compile time, the vector doesn't have to be constexpr for that

20:27

I wonder why std::array::size() can't be static aswell actually, that would solve this and be sane

Learath2

The size of the array inside the struct inside the vector is constant at compile time, the vector doesn't have to be constexpr for that

well but he is clearly accessing stuff, if he wants the size of the array, he can just constexpr the size of it

Jupstar ✪

well but he is clearly accessing stuff, if he wants the size of the array, he can just constexpr the size of it

hm, it wasn't that clear to me atleast

yeah actually nvm

20:29

just dont do std::size here, just declare the array size

20:29

can put a constexpr inside the struct probs?

20:29

so he can access it

Learath2

I wonder why std::array::size() can't be static aswell actually, that would solve this and be sane

constexpr auto k = decltype(decltype(v_vec)::value_type::m_arr)::size(); if this was the case, that'd be so nice

Jupstar ✪

can put a constexpr inside the struct probs?

Hm, that could work actually

mh no doesnt

anyway, the sizeof works too, so whatever

@Learath2 @Chairn ez pz no problem

c++

struct s{
    std::array<int, 100> a;
  

static constexpr std::size_t t = std::size(decltype(a)());
};

do you actually need the static too?

inside the struct yes

20:35

but should work outside of the struct too

Jupstar ✪

@Learath2 @Chairn ez pz no problem

c++

struct s{
    std::array<int, 100> a;
  

static constexpr std::size_t t = std::size(decltype(a)());
};

uh that doesn't work? Can't use a temporary inside std::size, that's what it told me when i tried

if its a constexpr why not

20:40

std::array constructor should be constexpr, but maybe only c++20 again xd

20:41

ah you probs using it on a datatype

20:41

maybe that datatype needs constexpr constructor too

error: temporary of non-literal type 'CGraphics_Threaded::SQuadContainer::SQuad' in a constant expression

yeah

20:43

SQuad constexpr constructor

20:43

gogog

this branch is becoming a huge mess

^^

i should split commits like Robyt does

20:44

i guess we're both extreme

what is this branch doing exactly :o

i bundle everything in a single commit, he breaks his pr in multiple small commits

Patiga

@Tater damn before I went away a week ago I opened your shadertoy page. now that I'm back that was the first page I have open and I'm absolutely amazed :D

Thanks :D

Voxel

what is this branch doing exactly :o

initialize everything everywhere

20:46

but it breaks the code right now (and im not sure, but im afraid it breaks protocol also...)

Chairn

initialize everything everywhere

is that even good btw?

20:46

that prevents catching logic errors that are caused if you didnt want a default 0 initialization

some part of the code are literally waiting to be broken to be honest

i imagine faster loading time

yes but i wouldnt just blindly say initializzing everything to default is a MUST be good

Voxel

i imagine faster loading time

should actually be slower, but in my preliminary tests, it was actually faster, i dont know why and how (edited)

santatrollet

1

Jupstar ✪

yes but i wouldnt just blindly say initializzing everything to default is a MUST be good

yeah i agree

20:48

im kinda in machine mode, not thinking

but anyway future coders could simply initialize values where its declared

20:48

struct a{ int i = 2 }

20:48

less error prone

20:48

;D

Jupstar ✪

but anyway future coders could simply initialize values where its declared

not even possible

but currently we also have constructors, oninit, onconsoleinit, onchillerdragonsmodneedssomeinitinit

because this auto declares constructor and can turn some classes into virtual...

20:49

which breaks all the mem_zero calls

20:49

i think i will revert all init and just fix the mem_zero, mem_copy and mem_comp calls

deen

what do you mean by focus it?

servelist is refreshing, when you switch tabs, but not when you come from ingome to server browser. is it worth to add a refresh there?

21:08

at least that's what i meant, that's the behavior for me, i'd expect to be the browser always up to date, when i focus it

21:09

coming from ingame, opening ingame browser, switching tabs, starting ddnet client, maybe even refreshing after a certain amount of time, when leaving the server browser open while afk or doing stuff (edited)

sizeof(*(int *)nullptr) isn't UB, right? (edited)

no idea, but i'd say it compiles and runs fine?

21:18

or does the standard require compiler to evaluate the expression inside sizeof, in that case it would segfault ? (edited)

heinrich5991

sizeof(*(int *)nullptr) isn't UB, right? (edited)

shouldn't be, evaluating that stmt is UB, but sizeof doesn't evaluate it's operand

21:28

well expression, not statement

Exported 158 message(s)