[CM] local variables not shadowing in defprocess

Rick Taube taube@uiuc.edu
Fri, 13 Aug 2004 08:39:16 -0500

> I'd add that lisp implementations have gotten worse, not better,
> in regards to the FFI's -- they're all different, and constantly
> changing.  And the free lisps (sbcl in particular) are being
> developed by people with what I regard as a C-mentality -- highly
> annoying to someone who remembers the good old days...

I have to echo Bill's sentiments about the dismal state of CLTL FFIs. I 
am currently trying to generate GKT2 FFIs for SBCL and OpenMCL so that 
CM can again have a plotters and browers (I actally have it working in 
openmcl now...) In addition to the coplexity of GTK, these two Lisp 
interfaces share basically nothing in common! OpenMCL makes it very 
easy to build an ffi (its just the output from a program called ffigen 
) but the restulting layer is clunky and very low level. SBCL's 
interface does a lot for you BUT you have to develop it by hand -- an 
impossible task for anything the size of GTK -- and its compiler is as 
strict as C even though Lisp is a dynamic language. Ive actually 
mananged to generate a valid gtk2 sbcl ffi  (only the typesystem so 
far) but it was more than two weeks of work to write the program that 
generated the FFI.

Its 2004 now and it shouldnt be this difficult to get a frigging cross 
platform GUI in CLTL.