Issue COMPUTE-APPLICABLE-METHODS Writeup

Issue:         COMPUTE-APPLICABLE-METHODS


References:    88-002R p.2-19


Category:      CHANGE


Edit history:  2-May-90, Version 1 by Moon


Problem description:


  COMPUTE-APPLICABLE-METHODS is documented as a regular function but it

  should be a generic function, because the metaobject protocol wants to

  allow different generic function classes to do it differently.


  This is Symbolics issue #30 and Loosemore issue #22 of 27 Feb 90.


Proposal (COMPUTE-APPLICABLE-METHODS:GENERIC):


  Change COMPUTE-APPLICABLE-METHODS from "Function" to "Standard Generic

  Function."


Test Case:


  (typep #'compute-applicable-methods 'generic-function) => t


Rationale:


  Needed for metaobject protocol.


Current practice:


  Symbolics Genera 8.0 and Lucid 4.0.0 Beta-1 implement the proposal,

  however the test case returns nil in Lucid due to an unrelated bug.

  Other CLOS implementations were not surveyed.


Cost to Implementors:


  Easy.


Cost to Users:


  None.


Cost of non-adoption:


  This will have to be changed later when the metaobject protocol is put in

  either as a standard or as an implementation-supplied extension.


Performance impact:


  None of consequence.


Benefits:


  Corrects a technical error in 88-002R.


Esthetics:


  Neutral.


Discussion:


  Gregor says this is correct.


  Originally Moon thought there were also problems with SLOT-BOUNDP,

  SLOT-EXISTS-P, and SLOT-MAKUNBOUND, but it turns out that those problems

  were corrected before 88-002R was published.