This site contains older material on Eiffel. For the main Eiffel page, see http://www.eiffel.com.

NICE Language Committee: ms-PROLOG

From Michael Schweitzer
Date: Sat, 16 Apr 94 15:06:08 +0200
To: bertrand@eiffel.com
Subject: Prolog
Cc: tynor@atlanta.twr.com
Content-Length: 645
X-Lines: 20
Status: RO

Dear Bertrand, dear Steve,

in separate mailings I am sending you a series of
questions and proposals. I realize this is quite
a chunk to swallow all at once, but on the other
hand I want to be sure I don't forget to send
off any of the questions on my mind at the moment.
Of course you're free to lay them aside for the
time being.

I believe that (together with my earlier postings)
these mailings cover all points that I find unclear
in the syntax and semantics of routines (in particular
instructions and expressions). I hope that all these
questions can be dealt with quickly in a manner satisfactory
to us all.

Best regards,
    Michael


From Michael Schweitzer
Date: Sat, 16 Apr 94 15:12:46 +0200
To: bertrand@eiffel.com
Subject: Epilog
Cc: tynor@atlanta.twr.com
Content-Length: 1170
X-Lines: 28
Status: RO

Dear Betrand, dear Steve

I think the questions I have posed and examples I have
given demonstrate that there are still some difficult
problems to be solved. Most of them have to do with
anchored types and expanded types. These concepts
are not complicated in themselves but when you combine
them with genericity the complications grow exponentially.
I believe we should all spend a few days thinking about
whether one or another of these bombs could best
be defused by restricting somewhat the use of anchored
and expanded types in connection with genericity.
After all, the usefulness of the examples I gave is
more than questionable; but as long as such things are
legal we must somehow come to grips with them.
Moreover, I ask you to bear in mind that some of these
combinations cause a significant runtime overhead
that can significantly reduce performance. Is it really
desirable to accept reduced performance just to support
some rather exotic constructions that will probably
almost never by used? After all, Eiffel will be judged
not only on the quality of the language but also on
the efficiency of the available implementations.

Best regards,
    Michael