Mirror of forum.nim-lang.org

691 :: FFI - cint vs int

[2014-12-27T22:32:55+01:00]

h42 (orginal) [2014-12-27T22:32:55+01:00] view original

I was looking through posix.nim and I saw that the 3rd parameter (buffer length) of "write" was "int" and the proc returned "int". I was wondering why "int" instead of "cint". I am also wondering if there is a general way to convert from int to cint. The manual says that cast just reinterprets the bit patterns, it doesn't actually do a conversion which could be a problem with 32 vs 64 bit integers. Thanks

Varriount (orginal) [2014-12-28T07:59:44+01:00] view original

In Nim, a cint is (almost?) always a 32 bit signed integer, whereas an int is an integer whose size is equal to that of a pointer (32 bits on x86 architectures, and 64 bits on x86_64 architectures).

If an int is used in the posix modules, then either the procedure signature varies depending on architecture, or the procedure signature is wrong.

Araq (orginal) [2014-12-28T10:05:27+01:00] view original

Usually int is used in the Posix module for size_t which matches pointer size in the same way int does.

h42 (orginal) [2014-12-28T16:16:55+01:00] view original

Thank you. I shouldn't have assumed I knew what int meant and read the manual more carefully.

PV (orginal) [2014-12-29T08:54:07+01:00] view original

@araq: that would be ptrdiff_t (signed type). size_t is unsigned.

Araq (orginal) [2014-12-29T10:08:11+01:00] view original

As if I don't know ... I don't like unsigned and mapping size_t to int is the way to go.