Hey John:

While I do not feel so bad about making it an error to assign an unbound
identifier, I do have to mention that this general attitude of feeling safe
to make anything an error because implementations can then do whatever they
want with it unsettles me a bit.  I think we are coming across quite
strongly when we say that it should be an error, no matter what the
implementation wants to do with that error.  If we really are neutral on the
view of what implementations should do and we do want to encourage an
underspecification, then we should use other language, such as calling it an
unspecified behavior or the like.  We should reserve calling things errors
when they may be considered legitimate.  Assuming that you twist things
around a little bit in your terminology, such as what Chez Scheme does with
calling all variables implicitly bound already, then you can get away with
saying that we really do not want to SET! unbound variables, and I think I
could generally agree with this, but it seems to become much a word game at
that point, and I wonder at that.

--
Aaron W. Hsu | arcfide@sacrideo.us | http://www.sacrideo.us
Programming is just another word for the lost art of thinking.

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports