[antlr-interest] Broken Features?

[Kay Röpke]

On Aug 17, 2008, at 1:10 AM, Foust wrote:

> Does it really make sense for Antlr to have such an expressive  
> syntax, when
> it is not fully implemented?
>
> rule :
> n+=('@'? NAME)                              // adds null to the  
> ArrayList
> n+= NAME                                          // works as expected
>
> There are many features like this, that are accepted by the grammar,  
> but
> don't work as expected and give no warning or other indication,  
> other than
> needing to be debugged.

Which ones? Sounds a bit like using ANTLR 3 is like walking a  
minefield, doesn't it?

> Is the problem design flaws that just don't have a
> good solution at the moment, or that v3, being a rewrite, is still  
> in the
> prototype stage (with the new syntax still being worked out)?

I'm sorry, somehow that sentence strikes as a bit non-constructive. I  
might be touchy, but I see problems like this one as minor issues not  
"design flaws".
I also strongly object to call ANTLR 3.1 as being in "prototype stage".
This question could easily have been framed as:
"I'm trying to do n+=(A? B) but it doesn't seem to work, instead it  
adds null to the list. Is this a known problem?"
instead of going the slightly more aggressive route.

> This particular feature (being able to assign an alias to an  
> expression) is
> really useful, but would be more so if it worked completely. Is the  
> intent
> to leave it partially implemented, fix it at some point, fix the  
> syntax to
> disallow unsupported expressions, or document the limitations and  
> produce
> warnings in the compiler?


http://antlr.org/wiki/label/ANTLR3/todo
http://antlr.org/wiki/display/ANTLR3/To+implement
and for this one:
http://antlr.org/wiki/display/ANTLR3/implement+labeled+set

And now that I look, I find both http://www.antlr.org:8888/browse/ANTLR-313 
  and the associated email thread from Jul/30-31.
Are you seriously expecting people (effectively Ter) to drop  
everything else and implement that in two weeks?

Of course, reporting bugs and shortcomings is fine and very much  
appreciated, don't get me wrong. After all that's what drives a project.

-k
-- 
Kay Röpke
http://classdump.org/