ptr::assimilate's name doesn't suggest what it does

[brson]

This function name may be too clever. Maybe to_unsafe_ptr

[catamorphism]

Well, we also have transmute, which isn't very obvious either... but, oh, fine. ;-)

[brson]

And pipes::entagle.

[nikomatsakis]

Although I suggested it's creation, perhaps assimilate() should just be converted to reinterpret_cast after all. Although it's "safe" to convert a region ptr to an unsafe ptr... it's only likely to occur in unsafe-ish code, I should imagine.

[jruderman]

perhaps assimilate() should just be converted to reinterpret_cast after all.

We should keep it. Type-restricted casts make code significantly easier to read and less prone to errors. With raw reinterpret_cast, the compiler will accept any of these, even though only one is correct. And I'm not even sure which one.

let a = ~7;
unsafe {
    let x : *uint = unsafe::reinterpret_cast(*a);
    let y : *uint = unsafe::reinterpret_cast(a);
    let z : *uint = unsafe::reinterpret_cast(&a);
}

With assimilate() (or to_unsafe_ptr), you know you're only changing the lifetime of the pointer.

[brson]

Maybe all of these can be named unsafe::something_cast for consistency.

[nikomatsakis]

maybe just cast_to_unsafe()

[bblum]

Is ptr::assimilate(x) not the same as ptr::addr_of(*x)?

[nikomatsakis]

It is, but I suppose ptr::addr_of will go away.