W3C home > Mailing lists > Public > xml-dist-app@w3.org > February 2002

Re: Reflective systems

From: Marwan Sabbouh <ms@mitre.org>
Date: Tue, 5 Feb 2002 13:16:56 -0500
Message-ID: <02be01c1ae71$4bb5aee0$253e5381@mitrezf21xy5nj>
To: "Kurt Cagle" <kurt@kurtcagle.net>, <xml-dist-app@w3.org>, "John J. Barton" <John_Barton@hpl.hp.com>
Sorry,  I think I phrased it wrong.  I did not mean the "is a necessary ..."  The point I was trying to make is: if you have the inheritance tree on the wire (see the ILU from Xerox PARC), it made it easy for me to implement reflections in a manner independent of the programming language.  nor did I imply that reflective is self descriptive or anything like that.  Thank you

Marwan
  ----- Original Message ----- 
  From: John J. Barton 
  To: Marwan Sabbouh ; Kurt Cagle ; xml-dist-app@w3.org 
  Sent: Tuesday, February 05, 2002 12:53 PM
  Subject: Re: Reflective systems


  Ok hold on here.  Reflective systems do not require
  strong typing: see for example Smalltalk:
  "Reflective Facilities in Smalltalk-80"
  Brian Foote, Ralph E. Johnson, OOPSLA 1989
  http://www.laputan.org/ref89/ref89.html
  or 
  Smalltalk: a Reflective Language
  Fred Rivard
  http://www.emn.fr/cs/object/biblio/publications/reflection96/reflection96.html
  (Smalltalk being the arch-typical *not* strong-typed language).

  And reflective is not self-descriptive.  "Reflective" is reasonably
  well defined in programming-language land.  Self descriptive is
  somewhat less well defined as I understand it, but most importantly
  the term refers messages not programming languages.  You can
  have a reflective distributed system without self-descriptive 
  messages and a non-reflective distributed system with 
  self-descriptive messages.

  I don't want to start a whole debate on this but neither should
  we be mixing apples and crankshafts.

  John.

  At 08:20 AM 2/5/2002 -0500, Marwan Sabbouh wrote:

    In my previuos research on middleware, I found that strong typing is a
    necessary condition for reflective systems.  Strong typing means that the
    inheritance tree is sent with every method invocation.  In this way, RDF may
    have much value to add beyond just interfaces.  Actually, RDF is at the
    heart of everything I am doing at the moment, and it is proving extremely
    useful.  ( I am hoping to get a paper out soon on RDF)

    Marwan

    ----- Original Message -----
    From: "Kurt Cagle" <kurt@kurtcagle.net>
    To: <xml-dist-app@w3.org>
    Sent: Tuesday, February 05, 2002 2:14 AM
    Subject: Re: Reflective systems


    > I'm jumping into this in the middle, so may be missing something, but I
    > would concur with the assessment that one primary disadvantage that binary
    > RPC mechanisms have is their inate inability to be self-descriptive. The
    > IDispatch mechanism works effectively if you have an IDL, but if you don't
    > then you're basically stuck in the dark about the characteristics of an
    > object (especially, as with many scripting interfaces, the core object
    only
    > acts as a proxy mechanism). The power of XML is not only that you can have
    a
    > descriptive interface, but you can in fact have any number of them - XSD,
    > RDF, WSDL, XTM, depending upon your requirements, and you can also utilize
    > multiple interface descriptors simultaneously.
    >
    > ----- Original Message -----
    > From: "Mark Baker" <distobj@acm.org>
    > To: "Rich Salz" <rsalz@zolera.com>
    > Cc: "Williams Stuart" <skw@hplb.hpl.hp.com>; <xml-dist-app@w3.org>
    > Sent: Monday, February 04, 2002 3:06 PM
    > Subject: Reflective systems
    >
    >
    > > >
    > > > > Absolutely true.  The wonderful thing about the Web is that this is
    > done
    > > > > *with* the Web.  It can describe itself.  RPC cannot.
    > > >
    > > > I disagree. Adding RDF to the web and saying it's reflexive seems no
    > > > different from adding DII to Corba, IDispatch to COM, etc.
    > >
    > > I don't know about IDispatch, but IIRC, DII did not reify interfaces as
    > > first class objects (i.e. they didn't get OIDs).
    > >
    > > Also, while that's necessary, it's not sufficient.  What's also needed
    > > is a uniform means of resolving an identifer.  Neither CORBA nor COM
    > > has that, but the Web has GET.
    > >
    > > MB
    > > --
    > > Mark Baker, Chief Science Officer, Planetfred, Inc.
    > > Ottawa, Ontario, CANADA.      mbaker@planetfred.com
    > > http://www.markbaker.ca  http://www.planetfred.com
    > >
    > >
    >
    >


  ______________________________________________________
  John J. Barton          email:  John_Barton@hpl.hp.com
  http://www.hpl.hp.com/personal/John_Barton/index.htm
  MS 1U-17  Hewlett-Packard Labs                                    
  1501 Page Mill Road              phone: (650)-236-2888 
  Palo Alto CA  94304-1126         FAX:   (650)-857-5100
Received on Tuesday, 5 February 2002 13:25:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:06 GMT