DDraceNetwork - developer

Learath2

https://github.com/teeworlds/teeworlds/blob/26d24ec061d44e6084b2d77a9b8a0a48e354eba6/src/game/client/components/hud.cpp#L641 0.7 one is much more concise, atleast imo

it also doesn't work correctly

00:01

the end pieces just disappear instead of getting clipped

don't they use tiny end pieces so that's not very obvious?

it's quite obvious once you see it

00:02

they just disappear xD

Ravie

it may just be clever maths to make the quad narrower while the texture stays in place

also from what I can tell this is the case

We really need better documentation for the graphics functions

00:07

We really need better documentation for everything actually

2

if you draw a pattern on the bar it becomes way more obvious how it's rendered

OHHHH

there could be a third sprite that gets drawn over top

Learath2

We really need better documentation for everything actually

please, prioritize the api

03:43

getting to this took me 5 hours

03:44

(this, for who's curious, is me logging the last 10 map finishes for me aka Sans3108 cuz i was trying to test a function that takes time.time and returns time.timeString) (edited)

03:46

probably should change the names of those but eh

Learath2

if you draw a pattern on the bar it becomes way more obvious how it's rendered

wtf 500 moving parts

Sans

getting to this took me 5 hours

javascript monkaS

03:53

you were saying?

03:54

my whole idea was to make a better ddnet bot, and to do that i need some sorta way to interact with the api, so i decided to make a wrapper for it and publish it on npm so i can use it in the bot KEKW

(edited)

03:54

and it's hard

03:55

i woulda used any other lang if it wasnt for the fact i only know to do stuff in js and also discord.py is outdated/doesnt cover slash commands (edited)

ddnet-maps

[ddnet-maps:master] 1 new commit

cc2fa2d A Disability - ddnet-maps

https://www.npmjs.com/package/ddnet

ddnet

Pulls data from ddnet.tw, otherwise known as "DDRace Network".. Latest version: 0.1.1, last published: 2 minutes ago. Start using ddnet in your project by running npm i ddnet. There are no other projects in the npm registry using ddnet.

06:19

nowhere near done

06:20

player part of things is almost done, possibly with bugs (i tried to check everything)

Hey, what are the oldest current solo r1 ranks? (edited)

Patiga

my struct RgbaSurface abstracts the srgb out of the chosen textureformat. if the format is non-srgb, it renders directly. if it is a srgb format, it renders to a texture and then to the surface

ah yeah u said that before, i remember, makes sense now

Patiga

I sadly don't have a geometry shader afaik, only vertex and fragment

doesn't matter anyway. Thinking about it for a bit I'd assume its 3 things that bottleneck in ddnet when zooming out far enough: 1. instacing is like a loop, so requires a powerful singlethread (as said grouping vertices together improves it slightly, so I assume drivers are already pretty optimized for instancing) 2. more load on the vertex shader. Even tho 2 quads next to each other instead of one big quad including both doesn't make a difference in fragment shader, it will create at least a small overhead in the vertex shader. So if you do it few thousands(or LOT more, more about that in 3.) it adds up to each other. 3. if you have a 1920x1080 screen, and watch BIT3 fully(without borders) one layer is already 1400x3900 tiles. For simplicity 3900x1400. I mean kinda obvious what i want to say for you. we often draw multiple quads for a single pixel. I assume 3. is indeed the main bottleneck in fact. The others might make the FPS fewer compared to taking your approach up to a certain zoom level, but its not like the vertex shader is totally slow, in the end you usually have less vertices than fragments. Except for 3.

I looked in renderdoc and see the insanity of overlapping quads

07:42

26 070 911 JUST for the top left border

07:42

my screen can only hold maximum of 3 686 400 pixels

07:43

impressive my GPU still renders it at 30fps xdd

07:43

that probs shows how optimized the rasterizer is for discarding overlapping pixels anyway

def-

[ddnet/ddnet] Pull request opened: #5604 Add 2 changes to formatting-revs so git blame is usable

With .git/config containing:

[blame]
  ignoreRevsFile = formatting-revs.txt

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 (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the cha...

going to move here since it's topical