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 2020-12-15 00:00:00Z and 2020-12-16 00:00:00Z

apparently trying to run 27 (or more) .cfg files back to back makes the game freeze and crash xD not sure if this is a good thing or bad

If it's reproducible please create an issue for it so it can be fixed

not sure why someone would have 27 cfg files linked to each other tbh, just a fun fact (edited)

yeah, but it's still a bug in the code, sometimes fixing one bug helps fix another

03:03

Though that one might just be your running out of stack space because we recurse too deep

also not sure if the amount of commands in each file makes any difference, I just have a bunch of files with this echo 21; exec meme_22.cfg; increasing each number with every file

https://www.youtube.com/watch?v=-RyhFF-PwYw client render "milk" map, but server running "Memories"

Definetly stack overflow

04:40

Just recursion

04:41

@Matodor cool! Reprodusable?

i think no

I entered the moment when the vote took place

[ddnet/ddnet] Issue opened: #3413 Client & Server map desync

https://www.youtube.com/watch?v=-RyhFF-PwYw By Matodor on Discord

lol, i have a demo, that too wrong map

04:49

uploaded to the pull request

@deen could you make a vote "Do you use english ddnet translation or other?" please? So I can see if it worth the effort to make "adaptive UI" or I can just go for optimisation and good english looking (edited)

05:17

Image attachment

05:18

I always consider possibilities of line be too long in other languages, but it looks bad if so

05:19

Example: ENG

Image attachment

05:20

RUS

Image attachment

05:20

for Ukrainian the "reset" word does not fit into button rectangle, but still usable

05:20

anyway looks ugly

05:21

if that worth it I could make this thing fully adaptive, so it will find the longest line and use the same offset for buttons, as well as adjust button width to fit the whole word

@Дядя Женя No need for a vote I think. Many use translations

Did I miss something in todays part 2? Where was the challenge? (edited)

08:54

Okay I guess you could have solved part 1 with a vector searching backwards for the last occurrence (edited)

11 hours of gameplay later, I can confidently say cyberpunk really feels like one of my projects. Overly ambitious but ultimately abandoned due to an exponentially increasing project complexity

Guys am I crazy, why does this segfault

#include <stdio.h>
#include <stdlib.h>

#define M 1000
#define N 1000
#define P 1000

int main() {
    double B[M][P];
    for (int m = 0; m < M; m++) {
        for (int p = 0; p < P; p++) {
            B[m][p] = rand() / (double)RAND_MAX;
        }
    }

    double C[P][N];
    for (int p = 0; p < P; p++) {
        for (int n = 0; n < N; n++) {
            C[p][n] = rand() / (double)RAND_MAX;
        }
    }

    printf("%f\n", B[0][0]);
    printf("%f\n", C[0][0]);
}

u probably run out of room o_o

10:19

try changing m n p to 10 or 100 instead and see if it still segv @timakro

wtf it works

10:20

but it's not that big is it?

16mb stack size

for the stack it's too big

is probably too big

10:22

ye

for heap no problem

just calloc

10:22

yeah lol

10:22

deen 2 fast

wow ok thx

Allocate more stack

[ddnet/ddnet-libs] Pull request closed: #15 Add libpng for Jupeyy (untested)

Learath2

Allocate more stack

how would u do that?

11:36

ulimit -s

11:36

i see

11:37

mine is 8mb

@timakro agreed, part 2 was no work today

11:51

I always get annoyed at the AoC guy misreading instructions for Part 1

I wished you could see part 2 from the beginning, I like to write my programs to print the answer to both so if I knew both puzzles I could factor out the common from the beginning

11:53

@deen Whenn will you start using some more exotic languages

I had some more exotic ones in the beginning

11:56

currently I've gotten lazy

right, i don't know what swipl is :D

Prolog

ah

Actually the best language for solving puzzles

12:02

I also wanted to use Z3, C++ Templates, C preprocessor, SQL, Makefile, Sed, but they are all a bit awkward to use

Makefile?

12:03

Can you program in a Makefile?

12:05

Apparently they are

12:05

turing complete

Yes, you can: https://github.com/kanaka/mal/tree/master/impls/make

mal - Make a Lisp. Contribute to kanaka/mal development by creating an account on GitHub.

12:05

Some other interesting language uses too

@deen If you have to kill some time ^^ https://codewithrockstar.com/

I'd rather use Ook!https://www.dangermouse.net/esoteric/ook.html

Haha, I like it :)

Lol what

13:07

Image attachment

Image attachment

there is a problem with the automapper in the latest version, the game crashes randomly when you use it

@TsFreddie yeah that's verybroken

13:11

The issue is that when the window is created the resolution is wrong

so on macos you have to set the GL viewport at start?

[ddnet/ddnet] Issue opened: #3414 Laser puller is not shown pulling in client anymore

As reported by Cøke, worked on 15.1.2 server, broken on 15.2.3

@TsFreddie is that on fullscreen or what?

Fullscreen on

13:25

toggle fullscreen off and on then I got the second screenshot

ok bcs we actually update the viewport in fullscreen and learath said in mac the canvas size is not equal to window size

13:26

so i could atleast try to patch that

@TsFreddie can u compile on that laptop?

13:45

or this https://github.com/Jupeyy/ddnet/runs/1557375770?check_suite_focus=true

@HMH you are alive!

13:48

️

13:49

sadSnail

13:49

rip

How to trick a client so that when entering a certain server, the map is loaded with my changed background.

You'd need code for that, since the client verifies hash sum

I don't know how to write code, is there a way for an amateur in this area to do this?

In CClient::LoadMap in client.cpp you could probably remove the hash-checking code

14:03

and then you need to recompile the client

[ddnet:master] 10 new commits

e40539c Settings Tabs Animations - Banana090 c02086a Fixes - Banana090 9b3d996 More UI Animation settings, more tabs animated - Banana090 b6d94a8 Visual Improvements, Fixed check-style - Banana090 b18d72d using time_get_mircoseconds and storing it as int64 val - Banana090 7e180f9 Fix rarest bug and typedef in UI animations - Banana090 119625d Merge branch 'master' into pr_animations_hud - Banana090 e234776 Merge branch 'master' into pr_animations_hud - Banana090 82faad3 Fix clang-format - def- 0f9f71d Merge #3409 - bors[bot]

14:05

[ddnet/ddnet] Pull request closed: #3409 Menu tabs animations

sneaky peaky, how he just evaded my requested changes by opening the same pr few months later

oooh

14:22

sorry, didn't remember that you had requests on the original change

@Deleted User

Image attachment

14:26

oof

14:26

sleep time, maybe next time

mhh looks annoying xD

well also doesnt matter probs @deen the initialization works bcs in tw everything is static, as soon as we wouldn't do that the zero initlaization would drop and it would require a proper constructor Aslong you read the pr for possible null ptrs its fine

14:35

and aslong its consistent and doesnt break after some time xD

I can't build on macOS again...

14:47

This is really driving me insane

cirBlech

If I get Command Line Tools I can't compile ddnet

i can't figure out why vscode isn't stopping at breakpoint

if I don't get command line tools brew can't compile python

14:52

This is all soooooo annoying, why do people like to work like this? Why do people tolerate these set ups? How do these even work for other people?

14:52

The state of windows and macos for development is just awful, awful, awful

14:52

How am I supposed to work when I have to mess around with my tools every 10 minutes?

14:54

Like now what? CMake is using the clang inside xcode, my developer directory is correctly set to the dev dir in xcode. Yet this clang is using the commandlinetools version of includes

14:54

Do I go digging into clangs source code?

Okay figured out it's cmake breaking, now what I dig into cmake?

have u used a clean cmake build dir

yes

has cmake no debug or verbose mode, maybe the path isnt set in the right order or smth

could we maybe get hook tiles on those platfoms here?

Image attachment

15:08

i see a lot of people blocking

15:08

on multeasymap

15:08

and i think that adding hook tiles on the floor would help people to defend

I narrowed it down to find_path finding a sqlite3.h inside commandlinetools, which is expected, but how it gets there I can't figure out at all

15:12

this is beside the point anyway, ofc I can debug it given an hour or two, I had an hour to spare and I wanted to look at the clipping issue @TsFreddie saw, now instead I'm wasting time wrestling with tools. Same experience on windows too

15:12

I never get to do any work, I only fix stupid tooling issues

@deen gonna create a quickpatch or a new version?

I bet this one is also going to boil down to some mistake in homebrew like every other macOS build issue I've had, because the person who made homebrew has a messiah complex and created a bloated mess

@Learath2 btw i tried to fix it: https://github.com/Jupeyy/ddnet/tree/pr_macos_fix_try

DDraceNetwork, a mod of Teeworlds. Contribute to Jupeyy/ddnet development by creating an account on GitHub.

15:14

u can download the artifact

15:14

and try if u want xd

I was surprised when I got the ddnet client building on windows in about an hour even though I never used visual studio before or compiled anything on windows

I'd just cherry-pick this one change? I'm not sure if we have other changes with high risk, but don't want to wait around for RCs

gj @timakro

ok

15:15

@timakro if u exclude the cmake process which downloads Gtest always on windows

15:15

then its strange

well ok an hour is some time but it can also take an hour alone to dl bloat studio

maybe disabled multiprocessor on VS?

It seems like there are a lot of options on windows with a separate git tool on the git website and the WSL git. Same for cmake, the cmake for windows tool works with a lot of IDEs. You can kinda mix and match and even if you don't really know what you're doing the tools most of the time work together. You never quite understand how though. They must have put a lot of work into making those tools compatible

@timakro you know what works great? Ignoring the windows way of doing things. As soon as I ditched the pipe dream that I'd get to do something the "windows" way once, I got it working in a matter of minutes

15:18

Just get msys2, works perfectly. Or use cmd, nmake/ninja and the console version of cmake, works perfectly

Yep

[ddnet/ddnet] Issue closed: #3398 Skin downloading produces graphics lag

15:18

[ddnet/ddnet] Pull request closed: #3402 Move PNG loading in SkinDL to fetch thread (fixes #3398)

deen

(this is for the Windows as a Linux distro argument: http://esr.ibiblio.org/?p=8764)

I really hope this guy is onto something

PES2_Depressed

windows

btw cmake has a lot of generators everywhere

15:20

I can generate sublime 3 projects, though no idea why anyone would want that

Meh sublime

15:21

Why would anyone use that

15:21

Vscode ecosystem is bigger

15:21

But vim is bettee

15:21

PES2_Depressed

Anyway, seems I will have to strace cmake to figure out what it's doing with pkg_config

15:22

I'll go have fun with that

[ddnet:15.2-devel] 3 new commits

383e744 Move PNG loading in SkinDL to fetch thread (fixes #3398) - def- b14e668 No need to call png_init every time - def- d800272 Version 15.2.4 - def-

"ld: warning: dylib (/usr/local/lib/libpng16.dylib) was built for newer macOS version (10.15) than being linked (10.9)" yeah now macos is onfire in the libpng pr too

Aaaaand ofc it's homebrew messing with things Variable declaration, 'homebrew_sdkroot' has value '/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk'

15:29

@Deleted User we either need to start shipping our own version of SDL or push SDL to finally fix the issue so we can start building for 10.15

15:30

(or we can build libpng too, we build our own curl and ffmpeg already)

i'd just ignore the system path comletly and dont do a prefer bundled, and then just static link everything

15:32

all problems solved for non linux OS

We do build our own libpng

yeah but not aggressivly enough xD

15:34

macos fails again

15:35

just rebased to the png_init commit

then just put some -DPREFER_BUNDLED_LIBS=ON in?

yeah maybe later, we also have to update the build scripts then etc. pp

15:37

@Learath2 yeah pls get SDL dev, and also fix the window focus under windows

I've completely lost all the motivation I had, I was going to fix a couple things today but I guess not

15:39

/Library/Developer/CommandLineTools/usr/bin/../include/c++/v1/cmath:317:9: error: no member named 'signbit' in the global namespace

look at this, what am I supposed to do about that?

15:39

Do I have to become a compiler engineer too?

do u use some experimental libc++ ? XD

No it's just CommandLineTools as shipped by apple

wtf xD

15:41

how does apple or the clang team even do that on macos i use experimental clang and had like no bug in 1 year xd

I remember I had the same error: https://stackoverflow.com/questions/58628377/catalina-c-using-cmath-headers-yield-error-no-member-named-signbit-in-th

Catalina C++: Using headers yield error: no member named 'signbit'...

After upgrading to Catalina from Mojave, Setuping: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.15.sdk in the env. I'm unable to compile a program t...

huh

@deen Yeah we had it on CI, I "fixed" it by rm -rf /Library/Developer/CommandLineTools

15:43

XCode comes with a complete set of tools already and they aren't broken

ok

cirBlech

there is no other os i know that goes so strong to push their own dev toolchain literally every macOS dev has xcode

Why ship 2 sets of tools even, why are these people like this?

2 sets?

Whenever I'm doing something I try to minimise work. Just isolate the common "tools" into a package and have the IDE use those?

15:47

CommandLineTools and XCode come with two entirely different set of "tools"

woot i did not know

I finally isolated the issue

15:47

no idea how to fix it though xD

so when i go to the appstore and dl 4gb xcode i get another version of the tools shipped in xcode-select --install ?

cmake is adding a -I/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk/usr/include to the compiler line, no idea what adds it, don't have any way to debug it

i like my way of dealing with these things

ChillerDragon yes

it's called giving up

hm

15:49

i thougth they depend on each other

15:49

i remember xcode-select --install caused a update popup for xcode

I have 4 copies of clang on my computer: inside XCode, inside the CLT, one I built myself to test something and another from brew

macports masterrace

15:50

why would you build clang ur self

15:50

u nuts?

That's the only way to rule out Apples insanity

15:51

I accidentally created a file called -c how do I delete it?

rm -- -c

thanks, I thought \-c would work but apparently not

it's not an escaping problem, the shell doesn't do anything with -, the program itself does

15:52

and it's a convention to stop interpreting -args after --

15:52

also important when you allow users to pass file names

15:53

and of course I know that because we have the maps -Ikarus-, -JustDoIt- & -SKY IVLAND- on DDNet

Okay I can't fix this, the includedir for sqlite3, curl and zlib are all /Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk/usr/include which is correct, but I can't have that passed as -I as includes that are part of the system are treated differently

15:55

For once I don't even have a vague clue what to do next

15:56

So I'll instead go sleep, what an absolute waste of time an entire hour and a half chasing a stupid issue because apple doesn't want to play nice with gnu

15:56

I'd honestly rather code on a limestone tablet

@Learath2 -isystem?

deen shell pro

Shell Pro Tip: Don't use shell scripts for anything public-facing

ima name my new ddnet map like a forkbomb ok?

call it "--no-preserve-root -rf /".map

xd

nah -isystem doesn't help either, nor could I possibly get cmake to use it anyway

16:02

hahaha /Library/Developer/CommandLineTools/usr/bin/../include/c++/v1/stdlib.h:20:15: fatal error: 'stdlib.h' file not found now I completely broke it

16:08

Oh I figured it out, finally

16:08

Still no idea how to fix it but for whatever reason cmake is using the macOS 11.0 sdk as the sysroot

16:09

So it's building with a very new clang, using a very old sdk, using the headers of a very new sdk

16:10

SQLite3_INCLUDEDIR:PATH=/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk/usr/include
SQLite3_INCLUDE_DIR:PATH=/Library/Developer/CommandLineTools/SDKs/MacOSX11.0.sdk/usr/include

Here riddle me this one, what could be the possible reason for this?

@deen Isn't the skin database powered by shell scripts?

yes, but not the public-facing part

Did I write this? https://github.com/ddnet/ddnet-web/tree/master/www/skins/edit

ddnet/ddnet-web

Static part of DDNet's website: https://ddnet.tw/. Contribute to ddnet/ddnet-web development by creating an account on GitHub.

only the moderator-facing part

That's comforting

16:12

I could have imagined past me hooking php up to shell scripts :D

16:13

I mean I hooked php up to python for trashmap

16:13

Ah it is accessible via web

16:13

You just need a login

16:13

I see

Think I found a bug in cmake

16:22

I was going to sleep, but noticed it's already 1700

16:27

Oh I actually should have went to sleep. I read 11.0 as 10.11

There is a simple application that wraps xcode so you can build for all apple platforms called xcrun

16:48

xcrun links to libxcselect.dylib which it immediately calls into and does nothing else

16:48

libxcselect.dylib upon this call loads another lib libxcrun.dylib dynamically and calls into that

16:49

which has many calls into libxcselect.dylib it's almost like italian bureucracy

heh, at the end of the chain, it execves an internal version of the binary xcrun

@timakro i like to call scripts using sudo from non admin user php code :D

ChillerDragon: you have set up sudo to only let through that particular command?

have you ever wondered about that dynamic libraries are quite bad concept?

17:40

i mean, dynamic libraries make sense in C, but in other languages, they are so awful to work with because maintaining ABI compatibility is hard

17:40

you usually end up with exporting plain C functions....

17:44

something interesting to read: https://www.kix.in/2008/06/19/an-alternative-to-shared-libraries/

hm. I like shared libraries

17:44

makes it easy to swap out stuff without recompiling everything

I found the oddest optimization I've seen in a decade easily

openssl security updates cough

This code is full of strcmp(s, "f") && strcmp(s, "find")

ehm, that doesn't sound like it's faster

does it do integer masks?

There is "r" "run", "n" "no-cache", "k" "kill-cache"

maybe they wanted ||

17:46

aaaah

just strcmp being strcmp

17:46

(and supporting < use case like that)

@heinrich5991 yes, but it's so fragile... even updating your compiler can break whole thing unless you export extern "C" things only

Image attachment

@Comrade that sounds like the problem of these languages not providing a stable ABI? ^^

yes, but it seems they just don't care

OOOH

this supports h or help

yep

17:47

It's so hard to think when I'm so sleepy :/

even modern things like Rust don't guarantee ABI stability as far as i know (edited)

wym, even C++ has some sort of abi stability

but only barely across compilers AFAIK

@Learath2 yes, but then someone tries to compile a single library on Windows using MSVC and RIP...

Yeah C++ is awful at it, that's why I gave it as an example

@Comrade I think there are initiatives to go for ABI stability in rust, but you're right, it doesn't exist right now

17:49

so you can't do some sort of rust shared library in a useful way

exactly

17:50

exporting C symbols is the only way

(except with a C ABI interface, that works, and works good in rust)

17:50

so why does it follow then that shared libraries are bad?

After an hour of reverse engineering xcrun, I finally figured out what's wrong

[6:40 PM] Comrade: have you ever wondered about that dynamic libraries are quite bad concept?

17:50

maybe they work badly with some languages in practice, but them being a bad concept?

17:51

with most*

17:51

@Learath2 what is it?

@heinrich5991 you think it's programming languages' fault?

xcrun says the current sdk is "10.15.6" since it consults _cltools_lookup_boot_system_sdk, but CommandLineTools ships with SDK 11.0, and cmake finds that and uses the latest SDK

17:52

So I end up with 10.15 includes and 11.0 includes mixing

17:54

At the end of the day it's apples fault for including sqlite3, curl and zlib headers in their own sdk, who does that even

17:55

Actually at the end of the day it's this entire ecosystems fault. homebrew uses the current sdk, xcode uses the current sdk, commandline tools uses the latest sdk, apple in their suggestions say you should use the latest sdk

17:56

cmake follows apples recommendation, homebrew just is a mess

@ChillerDragon I feel like there should be a way to run PHP as different users e.g. per application instead of everything running as www-data

@timakro well php-fpm which does the interpreting runs as a certain user

Yeah www-data ...

You could probably run multiple instances of php-fpm per application

Yeah could work, but I don't want to experiment with that and go against the "running everything as www-data approach" that's built into nginx, php-fpm and the distro

@Comrade well, I'd say "programming language X does not work well with shared libraries", not "shared libraries are a bad concept"

is there any programming language other than C that works well with shared libraries?

18:05

Pascal maybe?

idk. what about C#?

18:05

I'm not into that, so I don't know

Just because there is no good implementation doesn't mean the idea is bad

as far as i know you can easily load a DLL in C# and call some function from it, so probably yes

18:07

but does it work with non-MSVC DLLs?

I'd guess they follow some ABI, so I'd guess yes

I "fixed" the issue. My new build command is

cmake .. && sed -i "" 's/10\.15/11.0/g' CMakeCache.txt && ninja && install_name_tool -change "/usr/local/opt/sdl2/lib/libSDL2-2.0.0.dylib" "/Users/learath2/Desktop/C/SDL/build/libSDL2-2.0.0.dylib" ./DDNet

18:08

Overall amazing experience, I'm so glad I'm using this very mature and stable ecosystem

1

is macOS really that bad for devs?

I only had to patch every piece of the build process together

18:09

To be honest with you, the fact that I only had to launch 3 instances of ghidra suggests this was quite a simple issue to debug

if you need 3 instances of ghidra to debug our build process, maybe there's something wrong with it…

The install_name_tool one took 3-4 days, resulted in 2 SDL patches, and like 5 frameworks I had to reverse engineer

18:10

@heinrich5991 it's not our fault apple is out of their minds

@Learath2 I find your perseverance impressive

18:10

you're doing much better than I'd do

We could maybe include a heuristic in our CMakeLists to catch SDK mismatches?

oh no

sounds good

I mean idk I could file a cmake bug but is it really a bug? Apple suggests you use the latest sdk when building. It's homebrew that's hardcoding the stupid 10.15 sdk in their pkg-configs

how do we take the 10.15 from its pkg-config?

I could file a homebrew issue but they are kinda right too, xcrun does report the 10.15 sdk being correct because it refers to the boot sdk version

18:13

@heinrich5991 pkg_check_modules

18:14

it results in a pkg-config --variable=prefix <pkg>

18:15

https://github.com/Homebrew/brew/blob/master/Library/Homebrew/os/mac/pkgconfig/10.15/libcurl.pc

18:16

So from there I get the 10.15 sdk in my includedir, and from cmake internals I get 11.0 as the sysroot

18:20

I found a prettier way to do it SDKROOT=$(xcrun --show-sdk-path) cmake -GNinja ..

18:20

Was reading through the cmake darwin initialization thing and noticed SDKROOT overrides all the heuristics

18:22

I'll hold off on filing a bug anywhere

18:23

oh finally I can check the resizing thing

nice

Another interesting experience I had was when we were debugging the curl thing for osx cross compilation. I actually had to read a fair bit of clang source code for that one

18:28

@Deleted User not sure if it was you but we seem to have broken gfx_resizable somewhere along the way

18:29

I just remembered fullscreen was quite broken on macOS, @TsFreddie how did you even get it to launch?

18:29

Even 0.7 resorts to using "fake fullscreen" on macOS

18:31

If I launch ddnet fullscreen right now, it does some weird stuff then it starts windowed, yada yada we end up with the broken clipping, but changing resolutions fixes that quickly

Fullscreen is Broken on every OS (edited)

19:19

Useless function

huh

it's not useless, it can give more FPS

it never broke for me

I mean in OS API, not ddnet game (edited)

tee_thinking

how do you stop players from accidentally uploading their key material somewhere?

19:30

ideas: name the same something like "do_not_share.txt"

19:30

2) spread the keys across multiple files which would make it annoying to upload

19:31

3) use alternate streams, different positions to hide some key data

19:31

4) encrypt it with the local user profile (works on windows, probably mac, don't know about linux). chrome does this I think

19:32

5) encrypt it with some local details about the hardware, need some resilience in case the hardware changes

@heinrich5991 why are we stopiing people from uploading key material?

19:38

we don't have any key material

monkalaugh

ye, if we do add it at some point

19:38

I'm thinking accounts again

19:38

sorry

19:40

(hm, I shouldn't do emojis)

19:40

ah, I would put it in the keychain on mac, I bet there is a equivalent on windows, for linux we can try the gnome keyring, if not available we can try store it in an annoying way

yea, windows has something that chrome uses to encrypt the cookie db

19:41

(I know because chrome wouldn't let me borrow my cookies)

19:42

hm. but then it'd be hard for the user to transfer it to another PC

oh btw, I was reverse engineering LeagueClient (because it wouldn't let me log in without riot client), I learned about a very effective way to analyze TLS traffic decrypted

19:44

SSLKEYLOGFILE, did you know this existed???

19:44

Wireshark can directly parse this file and decrypt all the traffic

yes

19:45

it even works for the new quic libraries

19:45

at least for quiche (cloudflare's)

In CClient::LoadMap in client.cpp you could probably remove the hash-checking code

@deen what i need to delete in the file? Sorry, im not C++ progranmer.

Image attachment

enhance

Entering the danger zone writing C++ code when you're not a C++ programmer xD

@heinrich5991 if that's not available it seems your best bet is frida, you can just hook whichever ssl library they are using

YES

19:49

frida is amazing

19:49

I used it before

19:49

it's sooo good, and I only used a tiny bit of it

I think I advertised it here before

perhaps I got it from you, idk

Brilliant idea really, injecting a v8 engine into a program

hm, not sure how I feel about it being JS

19:50

but I guess it's the only real thing you can easily embed

19:50

so it's probably the right choice

19:51

@YupHio guess how I started in tw code

Well they could have injected a wasm vm too

less easy to handle I guess

19:51

you only want to script and suddenly you have to handle ABIs and whatnot

well V8 hooks about as well into C as WASM afaik

ehm I mean for the "end user" using frida

19:52

if it was WASM, you'd have to compile your frida modules

19:53

when most of the time, you'd be fine with some one-off script

ah they don't inject v8 apparently it's quickjs, could have sworn it was v8

19:53

anyway yeah I see what you mean

@heinrich5991 Haha same. Modding TW code is how I got into coding about 11 years ago, then I ended up going to college and got my degree in computer science. Now I'm a software developer, and all because I wanted to make an aimbot for a game that nobody played xD

@YupHio nice

20:22

where do you work now (if you're comfortable sharing)

I work at Boeing. It's not really a "software" company, but I love aerospace and they treat me well.

do you work on some high-assurance stuff? ^^

Yeah. Many of my applications are FAA accredited, and many are core/critical such that if there is a problem then airplanes potentially fall out of the sky...

20:29

Kinda makes it less stressful in a way, because the company is willing to give me as much time as I need to code/test things correctly, due to the high risk. Not the sort of job where we are pushing crap code out the door every two weeks lol

20:34

My applications also support a myriad of military programs... Which are arguably even higher stakes. I'm pretty luke-warm about being a part of that ethically, but anything that's cool and powerful enough inevitably gets used by the military.

20:37

The space stuff is awesome though. I have some applications that create and store data such as the parts and processes used to build and maintain the International Space Station, as well as other satellites. That's fun to think about. Supporting science but also supporting war feelsbadman

feelsbadman

YupHio

@heinrich5991 Haha same. Modding TW code is how I got into coding about 11 years ago, then I ended up going to college and got my degree in computer science. Now I'm a software developer, and all because I wanted to make an aimbot for a game that nobody played xD

A success story

20:42

Exact same story for me, but instead I've just kept doing tw stuff completely failing college

Lol

20:44

I failed a couple classes in college, first couple years I was just partying and chasing girls. Then realized that I actually had to graduate to get any ROI haha.

Atleast you partied. I just sat around for an entire year, followed by another year of sitting around, followed by another year of sitting around

20:46

Passed the bare minimum required to not be kicked out at the 2nd year

20:46

I also passed all computer related classes, so I only have boring engineering classes left...

Lol. I actually took a semester off from my university during my second year. Was pretty much failing everything so I withdrew and then changed my major to CS and came back in the Fall semester. Did the whole CS curriculum in 2.5 years and then I was out into industry

20:49

I was studying mechanical engineering for my first 1.5 years before taking a semester off. I was just too young to focus on school I think.

I should find a project where people would donate for my work and just quit uni. Just looking at this analysis book makes me pass out, there is no way I clean out 11 engineering classes this year

Dropping out for a semester ended up being a clutch move actually, because computer science is a much better fit for me. And I think CS actually pays slightly better than ME in most places

Learath2

I should find a project where people would donate for my work and just quit uni. Just looking at this analysis book makes me pass out, there is no way I clean out 11 engineering classes this year

if life was that easy

I don't really see anything I'd want to switch to tbf. I'm doing CEng at a very decent uni already. CS has even more discrete math than this (though I guess more useful math)

Ryozuki

if life was that easy

There is a guy I follow on twitter that is just developing linux kernel drivers for the new apple silicon, easily gets 5k$ on patreon

20:53

I'd kill to be paid to do reverse engineering work, or driver development

yeah there is always ppl that do x, then you try it and nobody sees you

Computer engineering is pretty cool if you like some hardware. Plus you can brag that you're a "real" engineer whereas computer science graduates are not technically "engineers" in the legal definition nor the industry definition.

Or the people making RPCS3, they make decent living wages making a ps3 emulator. How lucky do you need to be in life to get paid to make an emulator

i actually dropped out from computer engineering after 1 year, im instead doing a similar oficial thingy that gets me jobs anyway, i enjoy programming but uni rly made me want to die

20:55

sadSnail

But I remember taking statics and dynamics during my freshman year of engineering college. Those two classes were enough to scare away lots of students haha. Kind of a different way of thinking that you don't see often in CS but you see all the time in engineering

Hi, I think I need some help, I am out of ideas. So, I am searching the cause for a crashbug, which is very weird in my opinion. So the crash happens here: https://github.com/fokkonaut/F-DDrace/blob/F-DDrace/src/game/server/entities/character.cpp#L1463 I checked what exactly causes it and it seemed to be this line: https://github.com/fokkonaut/F-DDrace/commit/989e738de8e992649c4bffd703704c98047eadc5#diff-29ffe32916052e1066eed938021aadfeb29f855d7c63d9c1ca3350aceffb79f3R1465 I dont get how that can even happen, and the aBuf variables contains completely random data after crash, probably because it hasnt finished the formatting, but idk. The crash happens if a tee kills after a save reload, that means its a normal reload or shutdown and every player's tee is saved as a CSaveTee to a file, after reloading or server restart the tee gets its stats back. The crash doesnt happen consistently and never gave usefull information :/

literally the worst year of my life, idk what happened there to me, also i passed the programming related grades without batting an eye but math rly killed me xd

20:57

i couldnt see myself wasting 6 years of my life, cuz i was sure iwoudlnt make it in 4 yeears

fokkonaut

Hi, I think I need some help, I am out of ideas. So, I am searching the cause for a crashbug, which is very weird in my opinion. So the crash happens here: https://github.com/fokkonaut/F-DDrace/blob/F-DDrace/src/game/server/entities/character.cpp#L1463 I checked what exactly causes it and it seemed to be this line: https://github.com/fokkonaut/F-DDrace/commit/989e738de8e992649c4bffd703704c98047eadc5#diff-29ffe32916052e1066eed938021aadfeb29f855d7c63d9c1ca3350aceffb79f3R1465 I dont get how that can even happen, and the aBuf variables contains completely random data after crash, probably because it hasnt finished the formatting, but idk. The crash happens if a tee kills after a save reload, that means its a normal reload or shutdown and every player's tee is saved as a CSaveTee to a file, after reloading or server restart the tee gets its stats back. The crash doesnt happen consistently and never gave usefull information :/

I checked the BT, its a normal kill using the f1 kill cmd

its also expensive

I retook calc2 like 3 times due to goofing off haha. I can't believe how many math classes I took over the years at university....

xd

@fokkonaut can you check what of this is the invalid memory access?

i did

20:59

its the Server()->ClientName() as it seems

21:00

at least thats the line i am getting

you could do something like print Server()->ClientName() in the debugger

21:01

then it'll tell you if that is the invalid memory reference

I think this attitude in training computer scientists and engineers is stupid. I'm qualified to code, yet I probably will never get a diploma saying that, because I'm not an engineer in the 1970s sense

i dont know how, we would need @ChillerDragon for that @heinrich5991

damn interesting convo starts but i just got into a ranked doto game

21:02

@Learath2 same

21:03

i may not be able to develop a new super algorithm but i can code

we take real analysis 2 over here, I guess that's calculus 3 in the american sense

@Ryozuki Do what makes you happy, I think you made the right decision. Its all about having references on projects and code you can show anyway to find good paid jobs

yeah i also think i made the right decision

21:03

im at my 2nd year of the thing im making

Learath2

I think this attitude in training computer scientists and engineers is stupid. I'm qualified to code, yet I probably will never get a diploma saying that, because I'm not an engineer in the 1970s sense

Yeah, the whole 4-year degree thing was quite a lot of busy-work and jumping through hoops to "prove myself worthy" of the piece of paper that lets me get jobs lol. Lots of companies will hire you if you can code even without a degree, but it's hard to get your foot in the door especially at the big companies (like Boeing...)

next year ill be working

21:03

its more programming focused

21:04

i get top marks easily

21:04

im literally studying what the market wants rn

nice

web android etc

21:04

was a hard decision tho

YupHio

Yeah, the whole 4-year degree thing was quite a lot of busy-work and jumping through hoops to "prove myself worthy" of the piece of paper that lets me get jobs lol. Lots of companies will hire you if you can code even without a degree, but it's hard to get your foot in the door especially at the big companies (like Boeing...)

I wouldn't mind 4 years of computer science specific training. We are learning to be COMPUTER engineers, why do I need 2d projectile motion, I'm not becoming a siege engineer

cuz i was with the mindset that uni was everything too

21:05

its rly bad

Or fluid dynamics, god damn fluid dynamics, who decided all engineers need fluid dynamics

21:05

There are no fluids in a computer

water cooling!

21:05

@Learath2 u need to learn fluid dynamics to make a html website ofc

everywhere continuous non differentiable functions, just incase I want to make a fractal computer?

21:06

laplace transforms, I have never even once considered the laplace transform as an answer to any problem I've encountered

Lol. Most of the math and theory stuff I learned at uni is pretty much "good to know" aka I never use it. But it's kinda nice to know about it at least. I will never in my life have to solve integrals by hand, so maybe doing it in college was just to build character. Like how they make the kids dig holes in the movie "Holes"

21:09

No way around it. Had to endure the misery because the end outcome seemed so good.

math helped at the definitely not completely constructed day 13 problem

21:10

POGGIES

21:10

i need to retake aoc

21:10

been bussy

I mean I know about all the math. It's not enough to know "about" it to pass a real analysis class

21:10

I need to be able to prove all these weird things

Proofs are annoying

I wish I could bribe the dean to gimme a diploma

@Learath2 u just need to be rich and into politics

I think part of the reasoning is that they want people that can stand out from the crowd. You can argue "anyone" could code a simple android app, but how many people are willing to do frustrating busy-work for 4 years while not getting paid? xD that's the guy they want to hire.

a spanish politician did that

@YupHio I bet it has more to do with just tradition, but I guess that could be it too

@YupHio that sounds stupid. I see no reason why a company would have that particular reason

21:13

I can see the tradition argument

I spend days stepping through assembly, is that not perseverance enough?

21:14

They should add a second way to get all the credits. If you can identify and fix any bug in AppKit you get 180 credits

I think google has an alternative hiring process

Yeah. The 4-year degree is a big tradition thing. Shorter technical classes and certificates are becoming more popular because of that. But still, I think there is some value in the 4-year university experience. I definitely hated it when I was still working through it though. I think all engineering students hate it.

As I said I'd have no issue with a 4 year course on computer related stuff

21:16

There is one uni in berlin that seems to have a very modern and useful curriculum, it's the only one I've seen of the kind

which uni?

CODE

Many classes (at least in USA) allow you to test-out of the class if you already know the materials and don't want to do the busy work. Also, you can get credits for a class if you are doing related work as a hobby. Something to look into.

There are a lot of bootcamp style software engineer courses in france, etc. but this one is an actual uni

ah, but it's a private uni

ofc governments are where tradition lives

I hear arguments against the coding bootcamps sometimes. Something something you can't absorb it all in less than 4 years. Lol

do you know how much you have to pay?

no matter how progressive, governments profit from keeping the status quo

21:18

@heinrich5991 they have great payment plans, you only pay after you graduate

hm. do you have to pay interest? when is it due?

21:19

it would be bad if you'd have a lot of debt after studying

Oh apparently you can also pay it all up front, 32k euro

ok, doesn't sound too bad

I thought Germany abolished tuition? Or something like that?

32k euro for a bachelors is chump change compared to the US unis

@YupHio it's a private uni, they can charge. public unis don't charge (much)

Ah

it's 300€/semester here, and you get a ticket for public transport (but you can't opt out of that)

in italy I pay 240 euros a year

I went to a public uni (University of Idaho) and it was pretty cheap by USA standards. I actually graduated without any debt because I worked full-time in summers and almost full-time during the school year. Taking a semester off in the middle to work full-time also helped. Lots of money still lol

21:21

It was about $7,000 USD per year for tuition, plus room and board and books

7k$ per year sounds very decent for the US

21:23

Are you a US Citizen or is it subsidized for all students?

I got a cheap rate because I got "in-state" tuition since I lived in Idaho for more than a year before enrolling. If you enroll from out of state then it's about twice as expensive...

21:26

And yeah I'm a US citizen. I think it's actually cheaper for foreign exchange students because they are always trying to get more. Lots of different programs. I didn't qualify for any special programs or grants or scholarships, so I just paid the regular Idaho resident price.

ravie writing 3k lines

1

@Cøke and I found a little issue with tune zone messages, it's impossible to use quotes inside the messages because the message will end at the opening quote of the nested quotation. This makes it impossible to for example put a bind inside a tune message, for now the best we can do is use double apostrophes instead

have you tried \"

doesn't seem to work

@Ravie tune_message 1 "abc bind x \"+fire;+jump\""

21:38

idk what tune_message is called

yeah this seems to work

nice

for me at least, for coke the tune message behaves slightly differently on another version

that seems weird

21:40

what's the behavior for coke?

I'm able to just not put the whole message in quotes and it displays just fine, for him in that case it only does the first word

ah, that was changed recently IIRC

yeah thought so, but it still displays wrong for people with a slightly older client so it's not much good

no

21:42

it's the server version that matters

21:42

AFAIK

ah

21:43

well better use / anyway just to be safe since it works, thanks

ok work now, thx @heinrich5991

@Ravie it works without quotes since version 15.2

21:46

(not telling you to do anything, I just wanted to look it up)

https://www.iso.org/standard/79358.html

ISO/IEC 14882:2020

Programming languages — C++

21:49

they published c++20 standard today

huh, no hacker news post? ^^

Number of pages : 1853

21:50

aPES4_Steering

I'm paying them 0$ to keep me informed about such stuff

Now let's wait 5 years for any compiler to have meaningful support for it

@hear

21:51

@heinrich5991 why is the iso paid?

21:51

never heard of this

it's always paid

Image attachment

you can usually get the last draft version for free (which is essentially the same)

21:51

all the other C++ standards also cost a lot to read

is it a non profit or its actually greed?

Well I doubt ISO is non profit

https://en.wikipedia.org/wiki/International_Organization_for_Standardization

International Organization for Standardization

The International Organization for Standardization (ISO; ) is an international standard-setting body composed of representatives from various national standards organizations. Founded on 23 February 1947, the organization promotes worldwide proprietary, industrial, and commercial standards. It is headquartered in Geneva, Switzerland, and works ...

21:52

doesn't say non-profit, so I guess it's not

21:52

whether it's "greed" is probably subjective

21:52

but I don't like standards being behind paywalls, the ietf does much better here

idk why a standard would be paid

I don't know why ISO is involved at all

because people will pay and they like money? ^^

pepe_straight

21:53

well ill just use cppreference.com

21:53

sadSnail

I guess it gives your standard authority if it's published at ISO

21:53

yes, cppreference.com is reallly nice

21:54

still remember the times before that ^^

I also refer to the latest draft

https://en.cppreference.com/w/Cppreference:Archives

Cppreference:Archives

21:54

u can even download it

21:54

for offline use

but why does ISO give any extra authenticity?

love them

The C standard is ultimately written by wg14

well, because you convinced some standards people to publish your standard

21:55

btw

21:55

the quic rfc progressed to a state where they refer to version number 1

21:55

(no longer draft-29)

https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.10-Released

21:56

btw this too

21:56

5.10 big

21:56

"Linux 5.10 LTS Released As One Of The Biggest Kernel Releases In A While"

also some of the sub-rfcs are in "Draft is active, In: Waiting_for_Writeup"

21:56

maybe quic will be done soon™

Linux 5.10 LTS is likely the kernel to be used by Debian 11, Mageia 8, and others

21:57

POGGIES

Nintendo Switch controller support with Linux 5.10 thanks to the new Nintendo HID driver, another open-source community creation. Nintendo Switch Joy-Cons and Pro controllers should be working in both USB and Bluetooth modes under Linux as well as features like rumble support and gyro data. The driver previously has been available out-of-tree as a DKMS module.

21:59

ah finally

https://www.bbc.com/news/technology-55318225

EU reveals plan to regulate Big Tech

Facebook, Google and others face yearly checks and limits on what they can do with users' data.

21:59

i wonder

22:00

The two new laws involved - the Digital Services Act and the Digital Markets Act - have yet to be passed, so would only come into force after the Brexit transition period has ended.

22:00

DSA and DMA

22:00

DMCA?

22:00

xd

apparently you can say stuff like 'big tech' in a serious tone

xD

22:01

Facebook was one of the first to respond, saying it thought the laws were "on the right track to help preserve what is good about the internet".

22:01

i love companies answering this way

22:01

the irony

hm interesting. I would have expected a different response from facebook, regardless of the law

Apple declined to respond.

22:02

monkalaugh

22:02

@heinrich5991 https://github.com/google/mundane

Mundane is a Rust cryptography library backed by BoringSSL that is difficult to misuse, ergonomic, and performant (in that order). - google/mundane

22:02

did u know about this

I scanned today's hn front page, yes

im doing that now

didn't see the linux thing there

i found that on twitter

22:03

if u flw phoronix u can see lot of linux stuff

22:04

CMake 2.8.11 or later

22:04

sadSnail

ah, facebook moves uk's data out of the eu

ye

22:05

making it like california

22:05

or smth

Apple has always been great with privacy themselves

Kinda obvious what I'm saying: It shouldn't even matter where the data is, right?

The recent ios14 tracking prevention features are my favourite, every app needs to ask you before tracking

Exported 611 message(s)