[Openmcl-devel] Openmcl/SLIME problem

[Takehiko Abe]

Gary Byers wrote:

> This should actually work more reliably in the next release: if we get
> a number-of-args error on function entry, then we can't really act as
> if a stack frame has been built and arguments and local variables have
> well-defined locations in stack-frame slots or registers. 

I see. Thanks.

> One thing that -can- be done in 1.1 is to ignore/suppress errors
> that might occur when poking around in a partial stack frame. it'd
> be nicer to get an accurate idea of what the too many/too few args
> were in cases like this, but it's better to lie (return NIL) than
> to just blow up.

For Neil, guarding nbutlast forms seems to make slime happier.

e.g.

     (arg-vars (and vars (nbutlast vars nlocals)))
     (arg-indices (and map-indices (nbutlast map-indices nlocals))))