[cells-devel] Re: It's not a feature, it's a bug

[Thomas F. Burdick]

Kenny Tilton writes:

 > > > Thomas F. Burdick wrote:
 > > >
 > > > >  (defclass foo () ((a :accessor foo-a :type fixnum)))
 > > > >
 > > > >and if you try to set an A slot to something other than a fixnum,
 > > > >you'll get a runtime error. 

Okay, I need to lay of the crack a little.  PCL is not so well
integrated into SBCL as all this.  What I said applies to instances of
structure-class, when defined with defstruct, but not any kind of
anything you can define with defclass.  Oops.  It's still something
that should happen when PCL is better integrated, though.

Me again:
 > > > >-- or, have Cells propagate
 > > > >the type declaration to its new method definitions.  

I would still like to see this, though.  Hell, maybe cells:defmodel
will behave more in line with declarations=assertions than cl:defclass

 > If DEFMODEL stuffs type info by slot and class into some global hash or 
 > property list, at make-instance time we can pull the type info into a 
 > dedicated slot in the cell structure. would that help in anyway? Looks 
 > like THE cannot use a dynamic type.

Lemme think about this.  I bet we could have DEFMODEL construct stash
a reader/setter pair away somewhere and use them instead of
SLOT-VALUE.  There's gotta be some inoffensive way to leverage the
host CL's type system's abilities.

-- 
           /|_     .-----------------------.                        
         ,'  .\  / | No to Imperialist war |                        
     ,--'    _,'   | Wage class war!       |                        
    /       /      `-----------------------'                        
   (   -.  |                               
   |     ) |                               
  (`-.  '--.)                              
   `. )----'