Miss Dada  🏳️‍⚧️

Even if the function is declared as taking a pointer, references implicitly coerce, so it's rather moot what the decl uses IMO

Er, to clarify my tweet, my problem is not with using &, it's that the type being taken is fundamentally a pointer to a list of chars, not just one, and it would be nicer if a separate type existed for that, the way we have c_void

would @oli_obk's procedural vtable RFC address this?

it would if it allows for ZST DST fatness and also that DST pointers are repr(transparent) when the fatness is a ZST i'm not sure it does but iirc it does

I think it does allow that. here's the section on how it could be used for CStr https://github.com/oli-obk/rfcs/blob/procedural-vtables/text/0000-procedural-vtables.md#null-terminated-strings-stdfficstr …

You'd still need the signature to take that type though instead of a raw pointer

yeah, this doesn't obviate the need for CString, but it obviates the need for CString in footgun-prone APIs you control