>>>>> On Sat, 20 Sep 1997 16:23:47 -0500, Marilyn Bunzo <eos.arc.nasa.gov at msb> said:

> Has anyone observed class behavior whereby you can end up with more than
> one class defined of the same name in the same package?  We had been
> using some metaclasses which we initially blamed for the problm.  I
> eliminated the metaclasses from our system in the hopes of eliminating
> this error.  Unfortunately, that didn't work.  

[... snipped ...]

> We are using Allegro 4.3 on SGIs with all the latest patches downloaded
> from the Franz site.

> USER(7): (clos::class-direct-subclasses (find-class
> 'activity::contingent-activity))
> (#<STANDARD-CLASS ACTIVITY-SPACE::ND-AIR-MIDAS-DECIDE-ACTIVITY>
>  #<STANDARD-CLASS ACTIVITY-SPACE::ND-AIR-MIDAS-DECIDE-ACTIVITY @
> #x10864ada>
>  #<STANDARD-CLASS MIDAS-INTERNAL:DECIDE-BY-ALGORITHM>
>  #<STANDARD-CLASS MIDAS-INTERNAL:DECIDE-BY-ALGORITHM @ #x107c696a>
>  #<STANDARD-CLASS ACTIVITY-SPACE:DECIDE-BY-RULE>
>  #<STANDARD-CLASS ACTIVITY-SPACE:DECIDE-BY-RULE @ #x107b52da>)

I have observed this behavior when classes are re-defined.  It seems
that a new class object is created (in some cases?), which seems like
the only way to implement class re-definition in general.  I can not
recall any problems with this, other then confusion when doing things
like (clos::class-direct-subclasses (find-class <class>)).  I believe
that the objects without the address ("normal object identification
numbers") is the latest object and the others are old ones.

Art

-- 
Arthur Flatau                           Texas Microprocessor Division
<amd.com at Arthur.Flatau>                   Advanced Micro Devices
                                        5900 East Ben White Boulevard
                                        M/S 625 Austin TX 78741