From csr21 at cam.ac.uk  Fri Oct  1 15:23:37 2004
From: csr21 at cam.ac.uk (Christophe Rhodes)
Date: Fri, 01 Oct 2004 16:23:37 +0100
Subject: [Small-cl-src-discuss] Re: small mop compatability layer
References: <m2mzz6o4qi.fsf@bese.it>
Message-ID: <sqbrfm798m.fsf@cam.ac.uk>

marco <mb at bese.it> writes:

> The mopp package also exports the function
> SLOT-DEFINITION-DOCUMENTATION which while not strictly part of
> the MOP really should be and is implementened on most systems.

It surprises me slightly that you think that
SLOT-DEFINITION-DOCUMENTATION shouldn't be a generic function.

> IMPLEMENTATION - A keyword indetifying the implementation, one
> of: :OPENMCL, :SBCL, :CMU, :LISPWORKS.

> (let ((external-symbols '()))
>   (do-external-symbols (sym (find-package :it.bese.arnesi.mopp))
>     (push sym external-symbols))
>   (dolist (sym external-symbols)
>     (unless (provide-mopp-symbol sym #+openmcl :openmcl
>                                      #+sbcl :sbcl
>                                      #+cmu :cmu
>                                      #+lispworks :lispworks)
>       (warn "Unimplemented MOP symbol: ~S" sym))))

You could be slightly nicer to other implementations by doing
  ... (unless (provide-mopp-symbol sym (or #+openmcl :openmcl
                                           #+sbcl :sbcl
                                           #+cmu :cmu
                                           #+lispworks :lispworks
                                           :unimplemented))
        (warn ...))
and either provide a method on :unimplemented or a method on
NO-APPLICABLE-METHOD.

> (defmethod provide-mopp-symbol ((symbol symbol)
>                                 (implementation (eql :sbcl)))
>   (when (find-symbol (string symbol) :sb-pcl)
>     (import-to-mopp (find-symbol (string symbol) :sb-pcl))))

Unless there's a reason not to, you should be using the "SB-MOP"
package here.  (If there is a reason, sbcl-devel wants to know about
it!)

Cheers,

Christophe





From mb at bese.it  Fri Oct  1 16:27:30 2004
From: mb at bese.it (Marco Baringer)
Date: Fri, 01 Oct 2004 18:27:30 +0200
Subject: [Small-cl-src-discuss] Re: small mop compatability layer
In-Reply-To: <sqbrfm798m.fsf@cam.ac.uk> (Christophe Rhodes's message of
	"Fri, 01 Oct 2004 16:23:37 +0100")
References: <m2mzz6o4qi.fsf@bese.it> <sqbrfm798m.fsf@cam.ac.uk>
Message-ID: <m2655uo13h.fsf@bese.it>

Christophe Rhodes <csr21 at cam.ac.uk> writes:

> marco <mb at bese.it> writes:
>
>> The mopp package also exports the function
>> SLOT-DEFINITION-DOCUMENTATION which while not strictly part of
>> the MOP really should be and is implementened on most systems.
>
> It surprises me slightly that you think that
> SLOT-DEFINITION-DOCUMENTATION shouldn't be a generic function.

on openmcl and lispworks it is. on other lisps it's a wrapper around a
generic function, what's to be gained?

> You could be slightly nicer to other implementations by doing
>   ... (unless (provide-mopp-symbol sym (or #+openmcl :openmcl
>                                            #+sbcl :sbcl
>                                            #+cmu :cmu
>                                            #+lispworks :lispworks
>                                            :unimplemented))
>         (warn ...))
> and either provide a method on :unimplemented or a method on
> NO-APPLICABLE-METHOD.

good idea.

> Unless there's a reason not to, you should be using the "SB-MOP"
> package here.  (If there is a reason, sbcl-devel wants to know about
> it!)

the symbol metaobject exists in sb-pcl (though it does not name a
class), it doesn't exist in sb-mop. [hence the Cc to sbcl-devel].

-- 
-Marco
Ring the bells that still can ring.
Forget your perfect offering.
There is a crack in everything.
That's how the light gets in.
     -Leonard Cohen