[antlr-interest] ANTLR 3.0.1: invalid character column in a mismatch character error message.

[Kay Röpke]

On Aug 15, 2008, at 2:12 AM, Gerald Rosenberg wrote:

> Intended that the option value would also be used to generate a run- 
> time value as well.
>
> If only a BaseRecognizer modification is made, then the value would  
> not be available to the Tool during generation, right?  To be  
> consistent, the solution needs to cover development-time as well.

I will try to implement that over the weekend to see where it goes.  
Putting it into the runtime is a different story, but I've implemented  
the algorithm for a different project already, so I know it's possible  
to do in a subclass of Parser. It's more a question of "do we (and  
esp. Ter) want it in". I'd say it's a good addition, but that's not up  
to me to decide, naturally.

>> I'd hate another grammar option, btw.
>
> Agreed.  The alternative would be to provide a Tool level parameter  
> or method call, but that is not so plainly visible as a grammar  
> option.


The Tool version would look a bit different because ANTLR is actually  
implemented using ANTLR v2, so the mechanics would differ. I haven't  
looked at it too closely, but I know that for certain errors we don't  
have location info available (and I think it was because of v2  
limitations, but can't remember now).

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