A remark about research on reflection

Brian Rice water at tunes.org
Sat Apr 7 19:11:36 UTC 2001


>Hi
>
>I just to share some thoughts I have a about reflection.
>
>First I was thinking that research on reflection is ***not*** about redoing
>the past but I was wrong:
>People think that putting metaclasses in Java are interesting research:
>             http://home.austin.rr.com/forman/ira/jem.pdf
>(the conceptual contents of this paper is Smalltalk-76 -> ObjVlisp 87
>and some design ideas for Java).
>
>Second I really think that Smalltalk stopped its evolution in the area of
>reflection and other capabilities and it would be good if some new ideas and
>prototypes would be tested in Smalltalk and Squeak in particular.
>I think that I will restart working on reflection. I would really like
>to see if some other research groups would be interested to work on similar
>ideas.
>
>
>Stef

I recall reading Kent Beck's _Sorted Collection_ book some time ago, 
where he mentions some tentative experiments at factoring Object into 
a base and meta-object protocol, using the selector "meta" to get a 
meta-level protocol onto whatever object you're dealing with. This 
was intended to give a better conceptual factoring of the code, and 
to make it explicitly clear when someone was using the internal 
representation of objects or their implementation details. He 
probably has the code lying around somewhere, and I always have 
thought it would be interesting to see what Squeak would be like with 
that.

This would obviously be a heavy refactoring, but most of the major 
work would be in interjecting the word "meta" between object 
references and reflective message selectors. The crucial areas of 
course would be in Object and ProtoObject and such.

~





More information about the Squeak-dev mailing list