DDraceNetwork

archimede67

Are you on latest master

yep

same problem while building clear game

01:14

https://i.imgur.com/p7LBNWf.png

This does not look like latest master

01:25

@-StormAx looking at the commit hash (c91aa8807c79e7808ea81ce18ecdc259d80bf3be) this version is behind ddnet:master by 2 days

01:25

sync your repo with ddnet:master and see if you still have the error

archimede67

I had this bug which got fixed by #7575

Interesting, I didn't know it would already fix an issue. Invalid UTF-8 should never appear anywhere, so it being uninitialized memory makes sense.

-StormAx

fair, but i have no idea what u gonna do with it @furo

There should have been a .RTP file (the crash dump) in the same folder, which would tell you where the call to the function with invalid UTF-8 happened

09:15

If no .RTP files appear on crashes for you (on Windows), then crash reporting is broken for you

Robyt3

Interesting, I didn't know it would already fix an issue. Invalid UTF-8 should never appear anywhere, so it being uninitialized memory makes sense.

Yeah, for me it was the uninitialized m_aCmdPlayDemo that contained random characters that were indeed invalid UTF-8

this wouldn't have happened in rust, as you can't have uninit values or invalid utf8 in a string

this wouldn't have happenend in c++, as there is something called RAII

This wouldn't have happened in c, as reason

Fussel

this wouldn't have happenend in c++, as there is something called RAII

unrelated to the issue

13:49

also its in c++ the code kek

@Ryozuki there is a difference between c++ code and c++ compatible c/c++ mix

unrelated to the issue

btw, one could enable the compiler warning about uninitialized variables and explicitly disable it where it's intended, but warnings are "unrelated to the issue", i guess

did u ever check the ddnet code

13:53

anyway c++ doesnt ensure utf8 correctness anyway

13:54

in Rust, a &str is a valid utf8 always, the only way to contruct one would be using unsafe wrongly, and it would be UB to have invalid utf8 in itself

13:54

as for uninint, maybe if u follow RAII faithfully u may avoid uninit values, but c++ is made in a way it allows them quite easily (too easily) and they are bound to happen

it never silently happens if you enable compiler warnings and write destructors in a way leaving the objects in a valid state, that's exactly what i mean with the difference between c++ code and syntactically correct c/c++ mix

/usr/lib/ccache/bin/clang++ -DCONF_DEBUG -DCONF_INFORM_UPDATE -DCONF_OPENSSL -DCONF_VIDEORECORDER -DGAME_RELEASE_VERSION=\\\"17.4\\\" -DGLEW_STATIC -I/data2/edgar/ddnet/build/src -I/data2/edgar/ddnet/src -I/data2/edgar/ddnet/src/rust-bridge -g -std=c++17 -fdiagnostics-color=always -fcolor-diagnostics -fstack-protector-strong -fno-exceptions -Wall -Wextra -Wno-psabi -Wno-unused-parameter -Wno-missing-field-initializers -Wformat=2 -Wno-nullability-completeness -Wshadow-all -Wthread-safety -Wthread-safety-negative -Wsuggest-override -Wdynamic-class-memaccess -o CMakeFiles/engine-gfx.dir/src/engine/gfx/image_loader.cpp.o -c /data2/edgar/ddnet/src/engine/gfx/image_loader.cpp",
  "file": "/data2/edgar/ddnet/src/engine/gfx/image_loader.cpp",
  "output": "CMakeFiles/engine-gfx.dir/src/engine/gfx/image_loader.cpp.o

this is the warns we use rn

13:58

anyway c++ doesnt make it hard to do bad stuff, rust does

13:58

and people will always use a c++ mix

13:59

even more on a old game like this

people use kitchen knives to kill people, ban the kitchen knives ^^

bad analogy

13:59

i see ur a c++ maximalist

13:59

open ur mind! rust is better

»i see you are an atheist, open your mind to god«

im open to meet him anytime, its him who choses not to come owo

14:01

its sad, u should rly check rust, coming from c++ is where its most obvious how good it is

14:01

c++ place is only for legacy projects

Should have picked a cool 1 letter name for your language if you wanted people to think it's cool

rust name is cool, altho it atracts gamers, but its a shroom

14:02

i also think null terminated strings should die

Learath2

Should have picked a cool 1 letter name for your language if you wanted people to think it's cool

R

which is why my nasm no libc only syscall programs use strings with defined lengths

Ryozuki

i also think null terminated strings should die

I'm a null terminated string enjoyer

luckily the kernel doesnt enforce null terminated strings so far

14:03

syscalls*

ahh gg this isn't developer justatest

Learath2

I'm a null terminated string enjoyer

imagine needed to walk through the data to find its length when u can do it in o(1) (edited)

14:04

@Devinci type anyway

Imagine having to keep track of your strings legnth hypervomit

but u do XD

14:04

u do lots of STRLEN

Hello guyz this is bugs channel. To make it short @Ryozuki is right, was right and will always be right

Ryozuki

imagine needed to walk through the data to find its length when u can do it in o(1) (edited)

saving the lenght takes potentially not used space ig?

there are lot of optimization opportunities with having a known length beforehand

Ryozuki

imagine needed to walk through the data to find its length when u can do it in o(1) (edited)

For 99.999% of strings this is no issue, because they are short ^^

Devinci

saving the lenght takes potentially not used space ig?

not rly, its just u avoid needless work

Ryozuki

u do lots of STRLEN

Ackshully, I rarely use strlen, it's only ever needed when you need an algorithm that walks backwards or from both sides

null terminated strings make introducing bugs easy anyway

14:06

look all the string methods, we had problems with them

14:06

u need to ensure the null char

14:06

and that puts extra work to sometimes

Ryozuki

u need to ensure the null char

And for length prefixed strings you need to ensure the length is correct

u also put dev burden on to think about off by one errors