Bug in Victoria Day PCL (Lucid 3.0), using eql specializers

[Jon L White <jonl@lucid.com>]

This is a PCL bug (as suggested by Jordan K. Hubbard, who noted the
same problem in AKCL).  It doesn't happen with the "No Cute Name" release
version of PCL when using Lucid 3.0.

This is yet another instance of a common bug "leakage" in PCL; during the 
recomputation of effective methods for some altered generic function, a 
support function  -- in this case, COMPUTE-COLUMN-SPECIALIZERS from the
file points.lisp -- tries to call MAKE-INSTANCE, a generic function that 
just happens at that point in time to be "invalid".  From that point, it 
is in an push-down loop trying to update MAKE-INSTANCE by computing its 
effective methods, which of course necessitates calling MAKE-INSTANCE . . . 


-- JonL --