Re: flex default rule

Dimitri Maziuk <dmaziuk@bmrb.wisc.edu>
3 Apr 2006 01:41:22 -0400

          From comp.compilers

Related articles
flex default rule dmaziuk@bmrb.wisc.edu (Dimitri Maziuk) (2006-04-01)
Re: flex default rule johnmillaway@yahoo.com (John Millaway) (2006-04-03)
Re: flex default rule dmaziuk@bmrb.wisc.edu (Dimitri Maziuk) (2006-04-03)
Re: flex default rule cdodd@acm.org (Chris Dodd) (2006-04-03)
Re: flex default rule dmaziuk@bmrb.wisc.edu (Dimitri Maziuk) (2006-04-08)
| List of all articles for this month |

From: Dimitri Maziuk <dmaziuk@bmrb.wisc.edu>
Newsgroups: comp.compilers
Date: 3 Apr 2006 01:41:22 -0400
Organization: only myself
References: 06-04-001
Keywords: lex
Posted-Date: 03 Apr 2006 01:41:22 EDT

Dimitri Maziuk sez:
> Hi all,
>
> here's what I can't quite figure out:


Scratch that. I've been staring at my screen too long.


This is a backport from Java, and JFlex has a couple of differences
from flex. Most importantly, when more than one rule matches, JFlex
will apply the first rule in the set whereas flex applies the longest
matching one. As a result, I wrote the JFlex ruleset without regexp
negation and with only one lookahead. With flex I ended up using the
old "declare yet another lexical state and match .|\n in it" trick
in several places.


Dima
--
Relativity, Uncertainty, Incompleteness, Undecidability: choose any four



Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.