On Jan 17, 2010, at 5:46 PM, graham wideman wrote:
> I guess what's playing out here is the usual phenomenon that as  
> features are gradually built around a core codebase, lessons are  
> learned, and the legacy core model is more and more divergent from  
> some ideal core model for supporting the complete suite of  
> features.  So the incremental effort to add new features increases  
> in some exponential fashion, as each feature has to work around more  
> and more inconveniences of the legacy model.

Well put. that's where we're at.

> So, I think that you would end this sabbatical period a lot more  
> satisfied if you had completed a revamped enthusiasm-cultivating  
> core codebase than if you had fixed a few things, duct-taped on a  
> few new features, but were looking at a struggle for further feature  
> development.  In short, investing your undisturbed attention in the  
> new core model while you can sets you (and others including grad  
> students) up to be better able to add
> features incrementally after your sabbatical ends.

Yeah, that's my kneejerk reaction too.

> I do have a couple of questions:
>
> -- Do you expect the refactoring, or the further developments just  
> beyond that, to change the ANTLR language in backwards-incompatible  
> ways, or would we expect only additions to that language?

I would expect syntax/semantics to be same.  I'm hoping to leave  
runtime lib as-is.  Since it's a major version release, I might  
contemplate some new stuff.  Personally, the lexers still leave me a  
bit "cold".

> -- Might you attempt to absorb into ANTLR language (thus cross- 
> platform) some features that developers currently perform in  
> actions, for example building symbol tables?

An excellent idea.  'specially now that i have some well laid out  
patterns to use :)

> Thanks again for inviting me into this conversation.

you have an excellent perspective. thanks for joining in.
Ter
_______________________________________________
antlr-dev mailing list
[email protected]
http://www.antlr.org/mailman/listinfo/antlr-dev

Reply via email to