Improve the API for cstring - rust string conversion and vice-versa

[muzarski]

While we cannot make use of FFI trait API (we don't have any assumptions about the origin of the char* pointers provided by the user), we could probably utilize CassPtr to some extent. Thanks to that, we would be handling null pointers on a type level (represented as None).

I don't like current API for multiple reasons. One of them is that

pub unsafe fn ptr_to_cstr_n(ptr: *const c_char, size: size_t) -> Option<&'static str> {
    if ptr.is_null() {
        return None;
    }
    std::str::from_utf8(unsafe { std::slice::from_raw_parts(ptr as *const u8, size as usize) }).ok()
}

does not distinguish two scenarios:

1. pointer is null
2. provided bytes are non-utf8

In both of these scenarios, we return None.

[egorsavkin-plb]

std::str::from_utf8(unsafe { std::slice::from_raw_parts(ptr as *const u8, size as usize) }).ok()

Oh, I've run into this line of code crashing when I passed a string view to cass_row_get_column_by_name_n . Same call on original cpp driver is working fine.

[wprzytula]

std::str::from_utf8(unsafe { std::slice::from_raw_parts(ptr as *const u8, size as usize) }).ok()

Oh, I've run into this line of code crashing when I passed a string view to cass_row_get_column_by_name_n . Same call on original cpp driver is working fine.

That's worrying.

1. Can you provide the calling code?
2. How did the code crash? What did you observe?
3. Any ideas why this crashed here and didn't crash on original cpp driver?

[egorsavkin-plb]

I found out, it was a mistake on my side - I messed up with tying up the iterators and this was the consequence of it.

std::str::from_utf8(unsafe { std::slice::from_raw_parts(ptr as *const u8, size as usize) }).ok()

Oh, I've run into this line of code crashing when I passed a string view to cass_row_get_column_by_name_n . Same call on original cpp driver is working fine.