Remove null terminators from other extensions

[rasmusgo]

The field other in ExtensionSet currently handles null terminated strings in a weird way. It adds null terminators in fn names but doesn't remove them in fn from_properties. I got bit by this when checking for extensions by name and not having null terminators in my strings. I think it would make the most sense to not have null terminators in Strings.

[Ralith]

Good catch!

I think it would make the most sense to not have null terminators in Strings.

That makes sense to me. We'll need to insert nulls in fn names, then.

[rasmusgo]

Good catch!

I think it would make the most sense to not have null terminators in Strings.

That makes sense to me. We'll need to insert nulls in fn names, then.

It is already doing that.

[Ralith]

Revisiting this, there's a problem with the proposed approach: the most natural way to check for a name is to compare with xr::sys::WHATEVER_EXTENSION_NAME, which does include the null.

e: Maybe that's mitigated by those necessarily being the extensions that you can use the built-in bools for? Still feels inconsistent.

[rasmusgo]

The other field is used for extensions not available as booleans and those don't have constants either. The names of those are copied from documentation or by listing available extensions. It's also not really a problem to have redundant null terminators in the extension names when listing required extensions. What bit me was that I was trying to have optional extensions by comparing my strings without null terminators with available extensions with null terminators.

[rasmusgo]

The constants are also [u8] so I think it is fair that they contain \0 but the variables of type String doesn't.

[Ralith]

Fair enough.

[Ralith]

Thanks!