From mboxrd@z Thu Jan  1 00:00:00 1970
From: dewar@gnat.com
To: pfk@fuchs.offl.uni-jena.de, zack@codesourcery.com
Cc: gcc@gcc.gnu.org
Subject: Re: Proposal
Date: Wed, 19 Sep 2001 00:06:00 -0000
Message-id: <20010919070548.B0DC9F2B64@nile.gnat.com>
X-SW-Source: 2001-09/msg00749.html

<<You should investigate the rules for numeric syntax in languages
which already have this notation, such as Ada and Perl.
>>

In Ada, underscores can only be used to separate digits, and the extension
is thus a simple one from a syntactic definition point of view. The actual
excerpt from the Ada grammar is as follows:

2   decimal_literal ::= numeral [.numeral] [exponent]

3   numeral ::= digit {[underline] digit}

4   exponent ::= E [+] numeral | E - numeral

I don't see any language or definition issue in introducing exactly this
same restricted form into C


       16777216_UL
       16777216U_L
       0._1234
       0x_1234
       12.34_e+56
       12.34e_+56

All these should be illegal if the above approach is followed, and I think
that is the right choice. The only legitimate use of the underscore is to
separate digits in a long sequence of digits.