[antlr-interest] Two more lexer bugs in antlr-03-16-2007.10

[Gavin Lambert]

(Forwarding latest message to the list, since I haven't gotten a 
reply in this thread for a while.  Maybe the Internet gremlins are 
eating it.)

------ Forwarded Message ------
Date: Fri, 23 Mar 2007 20:44:26 +1200
To: Terence Parr <parrt at cs.usfca.edu>
From: Gavin Lambert <antlr at mirality.co.nz>
Subject: Re: [antlr-interest] Two more lexer bugs in 
antlr-03-16-2007.10

Yesterday, I wrote:
>If it helps, the following variation also demonstrates the 
>problem.  Since it avoids the use of inverse sets and uses unique 
>characters everywhere it might be simpler to keep track of.

Here's another one along similar lines.  The results are actually 
quite weird in this case -- rule D matches the single characters 
'1', '2', '3', and '5' and nothing else.  But again I guess it's 
the same root cause -- mistaking a sequence for a set.

Also, I've confirmed that all three test cases still fail in 
antlr-03-22-2007.18.  Has there been any progress yet?  Do you 
have enough to go on?

lexer grammar Test3;
fragment A : '1' | '2';
fragment B : '3' '4'+;
fragment C : '5' '6'*;
D : A | B | C;

------ End of Forward ------