Re: question about cdata type identification

[Francesco Abbate <francesco.bbt@...>]

2011/4/20 Mike Pall <mikelu-1104@mike.de>:
> The need to create and manage special predicate functions would
> complicate things too much.
>
> I've chosen to do it in the style of the other ffi.* functions that
> accept a ct (cdecl, ctype or template cdata) as the first argument.
>
> See: http://luajit.org/ext_ffi_api.html#ffi_istype
>
> And, yes, this function is JIT-compiled. The input types are usually
> constant or already checked, so this should be a zero-cost check
> in most use cases.

Hi Mike,

thank you very much, I really appreciate a lot what you are doing. For
me this problem was a showstopper to implement the generic arithmetic
operations so it is really helpful to have a timely response from you
on such a delicate API issue.

I know also that for you the focus in this moment is on the ARM
implementation so I appreciate even more.

I will keep the list updated about my progress for the FFI based
matrix module implementation with the GSL library. I believe there is
some interest for some of the list members.

Best regards,
-- 
Francesco