lagache@violet.berkeley.edu (Edouard Lagache) (05/01/88)
I don't have time for this, but as long as my train of thought is
derailed ..........
Perhaps I am back in fantasy mode again, but it seems to me that
the role of a standard is to codify existing practice to the extent
possible. Thus a naive starting point would be to simply do a
survey of all the existing code out there, and try to develop a
standard that maximizes the amount of existing applications that will
run under it.
Of course that won't work, and taking that approach too literally
results in monsters like Common LISP. However, that seems to me to be
the logical starting point, and I haven't heard that sort of noise from
the discussion on the BSI standard.
One very important reason for taking that perspective is to answer
the question of should a standard be implemented at all? While I don't
think this is a concern in this case, certainly it is true that a
language needs to mature before a standard makes sense. Another
question that can be answered in this way is what to standardize. At
the moment it appears that some of the BSI standard problems are
resulting from efforts to inject non-Edinburgh features into the
proposal. Perhaps things would be easier if one proposed to
standardize Edinburgh PROLOG only, and then worry about a larger
standard (I know, the non-Edinburgh folks wouldn't go for that since
that would leave them in the cold, but I am in fantasy mode, not
political mode!)
A last concern I have is over the question of language evolution.
Some aspects of the BSI standard seem to be directed toward making
PROLOG more "logical". I have stated this before, but it seems to me
that standards are not the way to achieve this goal. Rather I believe
that new languages are more suited to pushing back the frontiers of
logical purity. If someone can come up with a language which is as
usable as PROLOG and is better in those respects, then that new
language will eventually supplant PROLOG. In the mean time, PROLOG can
be standardized to support existing work, and work on these new
languages can continue without having their features frozen in a
standard at a time of relative "immaturity".
(I know, this is definitely in fantasy mode, but what do you
expect after spending 2 weeks chained to my computer writing
my thesis!)
Edouard Lagache
The PROLOG Forum
lagache@violet.berkeley.edu