Enough vapourware, here are the files.
ffi-patches and uffi.lisp
You'll see yourself that uffi.lisp is incomplete. Yet
it could use postgresql and (some) ODBC.
Please send all comments to clisp-devel *and* to me.
Maybe announce to clisp-users so they get a change to
try out this stuff?
Regards,
Jörg Höhle
Logged In: YES
user_id=377168
BTW, I think %cast1 can disappear in favour of a
foreign-variable constructor function.
Well, I should now talk less and use the few minutes left to
implement said stuff in foreign.d instead of playing with
clsql. People, please try it out! People, please try out
Pascal's UFFI layer!
Logged In: YES
user_id=377168
the updated uffi.lisp now does compile- (well,
macroexpand)-time instead of run-time UFFI->FFI
transformations and therefore pedantly checks the types.
clsql-3.0.4 is better used with this version.
Logged In: YES
user_id=377168
More macros added as I tested some UFFI code (browsing along
cells-gtk, clsql etc.) :pointer-self, :struct-pointer, setf
deref-array, def-union and many others are now supported.
Still, the deref-pointer / (c-ptr xyz) / pointer arithmetic
issue is problematic.
.asd file for ease of use with systems using ASDF
now allow compilation of ffi-patches-cvs.lisp
support cl-sdl and add some comments
Logged In: YES
user_id=377168
Added uffi.asd for convenience, but you must still load
ffi-patches-cvs before uffi!
However, now you could add this file to the asdf definition,
because ffi-patches-cvs can now be compiled.
Logged In: YES
user_id=5735
Originator: NO
Joerg, is there any hope that you will merge this into the trunk?
Logged In: YES
user_id=377168
Originator: YES
What do you mean with trunk?
a. the ffi-patches-cvs or
b. the whole UFFI package?
b. Could cause package problems, since CFFI includes an UFFI compatibility layer, which very likely defines an UFFI package also.
Logged In: YES
user_id=5735
Originator: NO
what I meant was "release": patching the clisp cvs head (and uffi) so that clisp supports uffi ootb.