Great thanks!
The method slove our problems.

On Wed, May 13, 2015 at 3:58 PM, Amaury Forgeot d'Arc <amaur...@gmail.com>
wrote:

> Hi,
>
> 2015-05-13 4:42 GMT+02:00 Yicong Huang <hengha....@gmail.com>:
>
>> However, the code is quite slow for reasons:
>> 1. Pypy treats the same function call for a new function every time. It
>> took time to do parsing, analysing and such a lot of work.
>> 2. JIT could not give any benifits for the single function call.
>>
>
> As said in the docs, the trick is to call pypy_execute_source once, and
> make it return a set of C function pointers.
>
> Here is how I would do it:
>
> - in a file "interface.py"
>
>     import cffi
>
>     ffi = cffi.FFI()
>     ffi.cdef('''
>     struct API {
>         double (*add_numbers)(double x, double y);
>     };
>     ''')
>
>     # Better define callbacks at module scope, it's important to keep this
> object alive.
>     @ffi.callback("double (double, double)")
>     def add_numbers(x, y):
>         return x + y
>
>     def fill_api(ptr):
>         api = ffi.cast("struct API*", ptr)
>         api.add_numbers = add_numbers
>
>
> - From C code:
>
>     struct API {
>         double (*add_numbers)(double x, double y);
>     };
>
>     API api;
>     const char source[] = "import interface;
> interface.fill_api(c_argument)";
>     res = pypy_execute_source_ptr(source, &api);
>
>     // Now call this in a loop:
>     api.add_numbers(12, 13);
>
>
> --
> Amaury Forgeot d'Arc
>
_______________________________________________
pypy-dev mailing list
pypy-dev@python.org
https://mail.python.org/mailman/listinfo/pypy-dev

Reply via email to