[jfli-users] jfli porting question

[Gary B. <gb...@cl...>]

Hi.

I've played around a bit with porting jfli to Clozure CL (CCL).  So far,
some things work and others don't; I assume that the things that don't
yet work are because of some mis-translation between FLIs/FFIs.

Specifically, when I try to reproduce the results indicated in the
"session.lisp" example and try to execute:

(setf p (jfli::new properties. :getproperty "fred" "ethel"))

I get a null pointer exception.

What seems to be happening is that JFLI::GET-SUPERCLASS-NAMES is
called recursively a few time on "java.util.Properties" and its
superclasses;
when it's called on "java.util.Map", the local variable "super" is bound
to a null pointer (not NIL).  That isn't expected, and we wind up calling
JFLI::CLASS.GETNAME on that null pointer.

This seems to suggest that JFLI::CLASS.GETSUPERCLASS is expected to
return NIL when its argument has no Java superclass (and a null pointer
would be returned at a lower level), but I can't seem to find (or at
least don't recognize) anything that's intended to arrange that.

Unfortunately, the only copy of LW that I have access to at the moment
segfaults before it reaches this point, so I can't easily check to see
how this is intended to behave.  Could someone who can check that (or
someone who's familiar with the code in general) please let me know
whether I'm correct in thinking that JFLI::CLASS.GETSUPERCLASS should
(for some reason ...) return NIL rather than a null pointer, or whether
the problem's likely to be elswehere ?

Thanks.

Gary Byers
gb...@cl...