Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!uwvax!rutgers!ames!ucbcad!ucbvax!computer-science.nottingham.ac.UK!rbh
From: rbh@computer-science.nottingham.ac.UK (Roger Henry)
Newsgroups: comp.lang.modula2
Subject: CanRead, Value and buffering.
Message-ID: <8708200907.AA01992@cayuga.cs.rochester.edu>
Date: Thu, 20-Aug-87 05:02:51 EDT
Article-I.D.: cayuga.8708200907.AA01992
Posted: Thu Aug 20 05:02:51 1987
Date-Received: Sat, 22-Aug-87 09:35:42 EDT
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: world
Organization: The ARPA Internet
Lines: 31

In message  <648@bingvaxu.cc.binghamton.edu>
Fred Sullivan writes:

> I am trying to figure out how to implement procedures which don't
> consume input, but may have to read far ahead, when the input is for
> example a terminal or a pipeline.  Good examples of what I mean are
> Int.CanRead and Int.Value.  Suppose the input is from a terminal, and
> the person sitting at the terminal types an endless stream of spaces.
> Should CanRead return FALSE after filling a buffer?  Should Value halt
> with an error after filling a buffer?  I can't think of anything else
> which would be reasonable, and I'm wondering what the proposers have in
> mind here.

The module BuffIn, in mailing 4 - yet to come, is designed for the
implementation of procedures such as Int.Value.  The specification implies
that non-significant characters (spaces) may be consumed - so the endless
stream of spaces will disappear.

==
Roger Henry
==
Address:
   Computer Science Department,
   University of Nottingham, Nottingham, NG7 2RD, England
Telephone:
   Nottingham (0602 or +44 602) 506101 extension 2855
Electronic Mail Addresses:
   JANET:       rbh@uk.ac.nott.cs
   ARPANET:     rbh%cs.nott.ac.uk@UCL-CS.ARPA
   UUCP:        rbh%cs.nott.ac.uk@ukc.uucp
   BITNET/EARN: rbh%NOTT.CS@AC.UK