Re: [PATCH] Factor %FLAG at scan level.

[Akim Demaille]

Le 10 avr. 09 à 20:35, Joel E. Denny a écrit :

> On Fri, 10 Apr 2009, Akim Demaille wrote:
>
> > > > > I would prefer:
> > > > >
> > > > > api.lex-symbol
> > > >
> > > > I really am not happy with "lex-symbol", but if you can't do  
> > > > better, I
> > > > certainly can't :)
> > >
> > > Other than the new prefix, I have nothing so far.
> >
> > Or maybe we should use the value, like you did for the api.  As you  
> > noted, it
> > doesn't make sense with other api related things, such as %pure.   
> > %define
> > lex.api symbol|pure|global?  Or api.lex.
>
> I like api.lex.  However, both "symbol" and "pure" are pure cases,  
> so the
> names don't convey the distinction.  Maybe:
>
>  %define api.lex-token "global|args|return"
>
> That is, we're specifying how lex communicates its token.

I like it.

> Although this
> looks reasonable to me, it's low-level.  It doesn't emphasize the
> high-level advantages. Maybe:
>
>  %define api.purity "impure|args|symbol"

In the case of yacc.c, purity is also about the interface of yyerror,  
and it has influence over the push-parser interface.  So departing  
from api.lex-token does make sense.

> So, it's the mechanism of purity.  Rather than saying  
> "global" ("global
> purity" sounds like a good thing),

:)

> we just emphasize that there isn't purity by saying "impure".

Good with me.  But you chose `args' over `arguments' and I'd like to  
stay away from abbrs.

> Maybe the best solution is a combination of these approaches.
>
> Anyway, it looks like api.pure in its current form will have a short  
> life
> span.

Yep.