Launching and crashing soon after about egregoria

[Uriopass]

As last time... Wayland+vulkan is buggy as hell with NVIDIA drivers. (Egregoria only supports vulkan on Linux as a graphics backend, I don't have time to make wgpu's opengl backend work)

When using the latest drivers I hit: gfx-rs/wgpu#4775
And when using the stable drivers I hit NVIDIA/egl-wayland#83 (comment)

🙃

Let's hope it gets better when the game is ready to be released! In the meantime, I can't test much.

[Uriopass]

But I'm mostly thinking it's wayland (which is networked) that's broken. In which case I can't investigate right now since I'd need to setup a wayland desktop.

[Arbee4ever]

I just remembered I can just switch to xorg by just logging out setting it to xorg and logging back in... Now it works fine :D

[Uriopass]

Wayland support is probably not great as I don't really tro to support it.
Last time I tried using wayland every other program was broken too so I thought I'd wait until it matured a bit.

[Uriopass]

Can you try running master instead? I upgraded to winit 0.29 recently which could help.

[Arbee4ever]

Tried that, and it didn't even launch and threw the error

thread 'main' panicked at 'texture not found at path: assets/sprites/blue_noise_512.png (in dir: Ok("/home/arbee/Documents/Egregoria")): The image format could not be determined': /home/arbee/Documents/Egregoria/engine/src/texture.rs:353
   0: <backtrace::capture::Backtrace as core::default::Default>::default
   1: log_panics::Config::install_panic_hook::{{closure}}
   2: std::panicking::rust_panic_with_hook
   3: std::panicking::begin_panic_handler::{{closure}}
   4: std::sys_common::backtrace::__rust_end_short_backtrace
   5: rust_begin_unwind
   6: core::panicking::panic_fmt
   7: engine::texture::TextureBuilder::from_path
   8: engine::framework::run::{{closure}}
   9: native_app::main
  10: std::sys_common::backtrace::__rust_begin_short_backtrace
  11: std::rt::lang_start::{{closure}}
  12: std::rt::lang_start_internal
  13: main
  14: __libc_start_call_main
  15: __libc_start_main_impl
  16: _start

even thought the texture at "assets/sprites/blue_noise_512.png" exists.

[Uriopass]

You probably didn't pull the LFS files, check the readme

[Arbee4ever]

I did run git lfs pull

[Uriopass]

What does
file assets/sprites/blue_noise_512.png
Show ?

[Arbee4ever]

This

[Uriopass]

Hmmmm, can you run again ? I don't see why the image format would not be recognized

[Arbee4ever]

Oh.. Is that a Terminal Command?
assets/sprites/blue_noise_512.png: PNG image data, 512 x 512, 8-bit/color RGBA, non-interlaced
Didn't know that command existed

[Uriopass]

It is :) very useful

[Arbee4ever]

I deleted the file and reran git lfs pull, and it ran, but I couldn't interact with it again until it crashed with

Io error: Broken pipe (os error 32)
thread 'main' panicked at 'Failed to run event loop: ExitFailure(1)': /home/arbee/Documents/Egregoria/engine/src/framework.rs:160
   0: <backtrace::capture::Backtrace as core::default::Default>::default
   1: log_panics::Config::install_panic_hook::{{closure}}
   2: std::panicking::rust_panic_with_hook
   3: std::panicking::begin_panic_handler::{{closure}}
   4: std::sys_common::backtrace::__rust_end_short_backtrace
   5: rust_begin_unwind
   6: core::panicking::panic_fmt
   7: core::result::unwrap_failed
   8: engine::framework::run::{{closure}}
   9: native_app::main
  10: std::sys_common::backtrace::__rust_begin_short_backtrace
  11: std::rt::lang_start::{{closure}}
  12: std::rt::lang_start_internal
  13: main
  14: __libc_start_call_main
  15: __libc_start_main_impl
  16: _start

I'm guessing that is the same error as before?

[Uriopass]

I have never seen this error before. Are you running any "surprising" setup? Maybe an old computer, a networked disk, maybe you don't have a gpu?...

[Arbee4ever]

Nope... I don't think so...

[Arbee4ever]

Maybe it's the nvidia gpu messing stuff up... wouldn't be the first time

[Uriopass]

I'm just wondering what pipe could be broken. Super weird.

[Arbee4ever]

I reran it yet again and got the error:

Io error: Connection reset by peer (os error 104)
Io error: Connection reset by peer (os error 104)
thread 'main' panicked at 'Failed to run event loop: ExitFailure(1)': /home/arbee/Documents/Egregoria/engine/src/framework.rs:160
   0: <backtrace::capture::Backtrace as core::default::Default>::default
   1: log_panics::Config::install_panic_hook::{{closure}}
   2: std::panicking::rust_panic_with_hook
   3: std::panicking::begin_panic_handler::{{closure}}
   4: std::sys_common::backtrace::__rust_end_short_backtrace
   5: rust_begin_unwind
   6: core::panicking::panic_fmt
   7: core::result::unwrap_failed
   8: engine::framework::run::{{closure}}
   9: native_app::main
  10: std::sys_common::backtrace::__rust_begin_short_backtrace
  11: std::rt::lang_start::{{closure}}
  12: std::rt::lang_start_internal
  13: main
  14: __libc_start_call_main
  15: __libc_start_main_impl
  16: _start

[Uriopass]

I'm thinking it might be related to the logging. Can you run

sed -i 's/release_max_level_info/release_max_level_off/g' native_app/Cargo.toml

then cargo run --release again?

[Arbee4ever]

I'm thinking it might be related to the logging. Can you run

sed -i 's/release_max_level_info/release_max_level_off/g' native_app/Cargo.toml

then cargo run --release again?

That crashed with Io error: Broken pipe (os error 32)
I might be able to try it on Windows sometime :D

[Uriopass]

Windows should definitely work. I run Ubuntu on X11 most days but I have a windows gaming pc where I try it occasionally and I haven't had any issues.

[Arbee4ever]

Yeah Windows can be more easily developed for because there isn't multiple display servers or something messing it up all the time

[Uriopass]

Nice! So it was a Wayland issue. I'll try to reproduce it as it's slowly becoming the default.

[Uriopass]

Goods news. I'm pretty sure this is fixed by gfx-rs/wgpu@e5c62fb which should come in the next wgpu release :)

[Uriopass]

Didn't made it in time for wgpu 0.19. It should be there in wgpu 0.20 though! (in a few months).