[antlr-interest] added attributes/dynamic scoping blog entry

Loring Craymer Loring.G.Craymer at jpl.nasa.gov
Thu Feb 3 18:00:14 PST 2005


At 05:43 PM 2/3/2005, Terence Parr wrote:

>On Feb 1, 2005, at 7:50 PM, John D. Mitchell wrote:
>>We need a parallel to init actions for exiting rules.  I.e., remember that
>>we can't presume GC or destructors. :-)
>>
>>Also, given that de-init sucks :-)...  Does exit work for people?
>>Personally, I don't like init/exit pair 'cause they don't match.
>>enter/leave seems the most consistent pair at the moment.
>
>I'm not opposed to an exit action...is this easy to code in C?  I guess so.
>
>So it's mainly for C and free() operations, right?

init() doesn't really fit with the other 3.0 improvements--especially when 
you have output grammars/templates to encapsulate most actions.  What would 
be useful is support for actions which set semantic state information 
coupled with a complementary set of rollback() actions.  These would 
operate in guessing mode and the state information would be used by 
semantic predicates.

--Loring


>Ter
>--
>CS Professor & Grad Director, University of San Francisco
>Creator, ANTLR Parser Generator, http://www.antlr.org
>Cofounder, http://www.jguru.com
>
>




More information about the antlr-interest mailing list