Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!iuvax!watmath!watdragon!djsalomon
From: djsalomon@watdragon.waterloo.edu (Daniel J. Salomon)
Newsgroups: comp.lang.modula2
Subject: Re: Subranges and doubledots
Message-ID: <14516@watdragon.waterloo.edu>
Date: 16 Jun 89 02:42:23 GMT
References: <983@maestro.htsa.aha.nl>
Reply-To: djsalomon@watdragon.waterloo.edu (Daniel J. Salomon)
Organization: U. of Waterloo, Ontario
Lines: 31

In article <983@maestro.htsa.aha.nl> fransvo@maestro.htsa.aha.nl (Frans van Otten) writes:
> In Modula-2, a valid subrange is [5 .. 10].  But would [5..10] also
> be a valid subrange ?  I mean, how should this be interpreted ?
>
>   [ 5 .. 10 ]   = "[" integer doubledot integer "]"
>
> or
>
>   [ 5. . 10 ]   = "[" real dot integer "]"
>
> What is the "common" interpretation ?
> -- 
> Frans van Otten                     |   fransvo@maestro.htsa.aha.nl    or
> Algemene Hogeschool Amsterdam       |   fransvo@htsa.uucp              or
> Technische en Maritieme Faculteit   |   [[...!]backbone!]htsa!fransvo

You have just encountered one of the rocky scanning/parsing problems of
Modula-2.  Pascal has a similar problem.

The sequence "[5..10]" is a valid subrange in Modula-2.  The reason is
that it can be generated by the Modula-2 grammar, whereas the sequence
"[ real dot integer ]" cannot.

The reason that it is a rocky problem is that you need two-character
lookahead in your scanner to recognize it.  A possible alternative is
to have your scanner recognize only integers and have your parser
assemble real constants from component integer fields.  Then your
scanner one needs only one-character lookahead.  If you use the second
method be sure to pass up all the characters of the integers rather
than just their values, otherwise the fractional part of a real number
may be interpreted incorrectly if it had leading zeros.