clisp ffi proposal: extend c-pointer

[Sam S. <sd...@gn...>]

Right now c-pointer as an argument to a function means a pointer to
foreign clisp storage.
I propose to extend it to also accept clisp arrays to mead the raw
pointer to their data vector (after accounting for displacement &c).
The datum would have to be pinned and handle_fault_range() would have
to be called.
I am reluctant to introduce a separate type name because I think the
same functions would want to accept c-pointers and pointers to lisp
data vectors.
This would enable users to do in FFI something similar to the rawsock module.

-- 
Sam Steingold <http://sds.podval.org>