[antlr-interest] v3.0 debug interface

[Andy Tripp]

>
>
>Ah, great, looks like we were on the same page as far as the inteface
>between GUI and lexer\parser goes, you just sent specs in terms of Java
>interface rather than the net based protocol. And I certainly wasn't
>think of some nasty CORBA style thing, just a text protocol like you
>suggest.
>
Just thought I'd mention that I really like CORBA, and I think the days 
of everyone inventing
their own little text-based protocol across a socket are over. However, 
in this case, it
may be overkill, as the complexity of the data is very low.

>
>As to the format of the interface, I meant the standard Java style
>listeners, extend java.util.EventListener (just a tag for tools) and
>then (add|remove)Listener(...); methods (hadn't remembered that Java's
>"Listener" pattern was quite so simple, too long living day to day in
>C#'s delegate land I guess).
>
>Amd then events of the form:
>void location(LocationEvent evt); 
>where LocationEvent extends java.util.EventObject.
>
I'd certainly recommend using Listeners (I assume that's what Terrence 
had in mind).
But I wouldn't extend any of the built-in Java Events, as you don't need 
any of
that functionality.

>
>While I agree about the general issue of cost\benefit, these changes
>seem of little cost, seem to offer some functional benefit (the two
>use-cases of modular handling in the GUI and code-gen specific events)
>and also fit the standard Java design patterns thus making them easier
>to understand for those familiar with other Java apps (in this case a
>rather weak point due to the simplicity of the pattern but still) and
>understandable by other language tools.
>
Also, extending built-in Java Events will make handling them in other 
languages a mess.

Andy