demonstrate incorrect usage of RelPath::new_unchecked

[BenjaminBrienen]

No description provided.

[ChayimFriedman2]

I do not object to this specifically, but I don't want to start documenting internal APIs. Keep that in mind.

[Wilfred]

The fact that unit tests are failing with this new assertion suggests that perhaps it's worthwhile? It looks like it's trivial stuff where we have e.g. //- /submod.rs in tests but being extra defensive around any foo_unchecked seems nice to me.

[Veykril]

odd this does not fail on windows 🤔

It looks like it's trivial stuff where we have e.g. //- /submod.rs in tests but being extra defensive around any foo_unchecked seems nice to me.

its not those, otherwise almost every test ought to fail. It seems to be a bug in our strip_prefix implementation?

[Veykril]

Ah I think this happens if you strip_prefix with an empty path arg?

Yea, and the fact that our VirtualPath implements pop incorrectly. It turns a rooted path into an empty path, thats also a bug

[BenjaminBrienen]

Just close this when you feel like you don't need the demo anymore