Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site whuxle.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!ihnp4!whuxle!mp
From: mp@whuxle.UUCP (Mark Plotnick)
Newsgroups: net.unix-wizards
Subject: Re: Comment recognition in Lex, again
Message-ID: <340@whuxle.UUCP>
Date: Sat, 5-May-84 13:54:38 EDT
Article-I.D.: whuxle.340
Posted: Sat May  5 13:54:38 1984
Date-Received: Sun, 6-May-84 01:31:24 EDT
References: <245@uwvax.ARPA> <1876@utcsstat.UUCP>
Organization: Bell Labs, Whippany
Lines: 42

The problem with
	"/*"([^*]|("*"/[^/]))*"*/"
is that the right context handling in lex in nested regular expressions
is a little nonintuitive.  After lex recognizes the complete
expression, it backs up one character because of the ``/[^/]''
expression.

In case you still don't see the problem, run this lex program:

	a(b/c)c { printf("I saw this: %s\n", yytext); }
	.	{ printf("char: '%c'\n", yytext[0]);

The first rule will NOT match ``abc'', but it will match ``abcc'',
sort of.  It prints out ``I saw this: ab''.

To be safe, only use right context at the very end of your regular
expression.

Yet Another Way To Recognize Comments:

I really don't enjoy beating my head against a wall playing
with regular expressions and starting conditions.  When we
had to write a compiler a couple of years ago (any other
AM295 survivors out there?), we did something like:
"/*" {
#define LEXEOF 0
	int c, last_c='\0';
	while ((c=input()) != LEXEOF) {
		if (last_c == '*' && c=='/')
			break;
		else
			last_c=c;
	}
	printf("comment seen\n");
	if (c == LEXEOF)
		printf("EOF within comment\n");
    }

Moving some of the effort into the action routine allows you to easily
add more context-dependent features, such as printing a warning message
if there's a ';' within the comment, supporting nested comments, etc.
	Mark Plotnick