bug: SVG randomly stops working in Linux

[marc2332]

For whatever reason when I open a folder in freya-editor, all SVGs stop working

[marc2332]

[marc2332]

I wonder if it could be because of

https://github.com/marc2332/freya/blob/9d37c3c2532c680b309d35bdb5561bc91be26400/crates/renderer/src/window.rs#L63-L64

[marc2332]

Hey @pragmatrix, sorry for pinging you here, do you have any idea what could be causin this? It doesn't happen on Windows.

Maybe #52 (comment) ? I haven't tried disabling it yet, will do soon, but I doubt that's the issue..

[marc2332]

For reference, this is how it should look:

[pragmatrix]

I've no idea. Initially I though it may be that opening a folder may change the current directory on Linux, but as I've seen in the source, the svgs are loaded at compile time.

[marc2332]

Yeah, they are statically linked, I wonder if it has anything to do with loading emojis (I am using those for the folder and file icons), will make some experiments soon

[marc2332]

after some testing

• not related to emojis
• not related to bug: SVG randomly stops working in Linux #52 (comment)

mmm

[marc2332]

For whatever reason when I open a folder in freya-editor, all SVGs stop working

Actually, it also happens in other situations, I don't see any pattern

[pragmatrix]

If you are confident that the problem is in the Skia part, I suggest to compile the project with SKIA_DEBUG=1. May be some assertions will show up in the terminal output when the SVGs fail to render.

[marc2332]

If you are confident that the problem is in the Skia part, I suggest to compile the project with SKIA_DEBUG=1. May be some assertions will show up in the terminal output when the SVGs fail to render.

I tried to compile it on the CI but I couldn't get it to work

And now I tried to compile it on my linux but got SIGSEGV: child process 'cargo' core dumped when finally running the binary, any other way I could do it?

[marc2332]

Looks like it doesn't crash with SKIA_DEBUG = 0 😭

[marc2332]

Any idea why is it crashing with SKIA_DEBUG = 1? @pragmatrix

[pragmatrix]

No, but the next step would be to take a closer look with a debugger or analyze the crash dump.