[Cython] Semantics of cdef int division and %?

[Stefan Behnel]

Sturla Molden wrote:
>> There is a very clear distinction between all types, including Python
>> types
>> and C types. The internals are not always simple, but the syntax is and
>> the semantics tries to be, which is the thing that matters.
> 
> You are adding Python syntax to some but not all C types

I assume you meant semantics here.


> with the
> argument that you are trying to make Cython an optionally statically typed
> Python.

It's actually more complex than that. There are constraints that come from
the compatibility to Py2.x and Py3, from language features in C and Python,
from different C standards and compilers, ...

Once all of that's been considered, the remaining design decisions are
rarely black-or-white. This very thread is just one example.


> So 'cdef int' gives the behaviour of a Python int, whilst 'cdef
> float' gives the behaviour of a C float.

This is wrong, please read what I posted already.


>> This discussion is not about right or wrong, it's about good semantics
>> that matches the expectations that users have of the language.
> 
> What does we expect
> 
>    cdef float a
> 
> to mean? Single or double precision?

"float" is a C type. You do the math.

Stefan