johnl@haddock.UUCP (05/08/84)
#R:watmath:-766600:haddock:16800012:000:685 haddock!johnl May 7 14:28:00 1984 The problem with recognizing comments as single Lex tokens is that Lex has a finite sized token buffer. If your comments are bigger than the buffer, you lose. What should work is to put in a COMMENT start state and do this: "/*" BEGIN COMMENT; /* start a comment */ <COMMENT>"*/" BEGIN 0; /* return to regular state */ <COMMENT>.|\n ; /* throw everything else away */ Unfortunately, I've never had much luck with making lex programs that include start states work. Everybody lets lex recognize the /* and writes C code to eat the comment; if you read the manual page for lex, you'll find exactly that in the example. John Levine, ima!johnl
smk@axiom.UUCP (Steven M. Kramer) (05/11/84)
I had used lex a few years ago and needed (actually wanted) to implement
comments. I made a special comment state to do it in. Otherwise, if
you try to match a comment in 1 rule, large comments will blow away the
buffer. I had no trouble in making a separate state for comments.
Where's the beef?
--
--steve kramer
{allegra,genrad,ihnp4,utzoo,philabs,uw-beaver}!linus!axiom!smk (UUCP)
linus!axiom!smk@mitre-bedford (MIL)dan@haddock.UUCP (05/23/84)
#R:sdcrdcf:-109200:haddock:16800013:000:222 haddock!dan May 22 15:32:00 1984 Or rather, don't bother preparing for comments like those. The line containing the word "variants" is incorrect, as you can see by running the message through the C preprocessor. (It produces an extraneous close-comment.)