[Ecls-list] updating refererence pointers in UFFI

[Goffioul Michael]

> This is not only a CLISP feature, SBCL and CMUCL also have 
> it. This should be done at the level of the c-inline special 
> form. It will indeed complicate the special form a little, 
> given the fact that c-inline already supports multiple values 
> as output type.
> 
> I am worried however at messing up with the UFFI interface, 
> but that can probably be fixed with a warning message and a 
> global flag that disables it. As a side note, if you look at 
> all the FFI interfaces out there, not even CFFI provide this, 
> probably because some commercial lisps do not support in/out 
> and out arguments.

If you consider that any argument without a in/out tag is an
input argument, then you should be compliant with UFFI and the
user should then allocate the foreign objects himself and
dereferencing it in the UFFI-way. But IMO, UFFI/CFFI should
integrate those concepts because it makes foreign programming
more natural (LISP-like).

Michael.