On Mon, Apr 17, 2006 at 07:25:27PM +0100, Nicholas Clark wrote:
> On Mon, Apr 17, 2006 at 10:15:40AM -0700, [EMAIL PROTECTED] wrote:
> > * grammar_rules.pge - support for POD sections.
> > ## This rule handles whitespace and comments between tokens.
> > -## XXX: Add pod directive handling
> >
> > -rule ws { [ \# \N+ | \s+ ]* ::: }
> > +rule ws {
> > + [ \# \N+
> > + | \s+
> > + | ^^ =cut \s* $$
> > + | ^^ =for \s* \N+ .*? ^^$$
> > + | ^^ =\w\N* .*? ^^ =cut $$
> > + ]* :::
> > +}
>
> [People not reading the commit list should note that this wasn't the final
> form]
>
> I see you've taken out the XXX. However, isn't the plan for Perl 6 that all
> the POD is going to be available as <DATA>-like streams? Whereas whitespace
> does get eaten. So surely they should be 2 different rules?
Since we can embed closures and subroutines in rules, we can capture
the POD as we eat it. :-) In the long-run it will likely end up as:
rule ws {
[ \# \N+
| \s+
| \. \# <PGE::Text::bracketed: <({[>
| ^^ <pod>
]* :::
where the <pod> subrule can keep track of the appropriate things
to make the <DATA>-like streams.
Audrey originally had some of the pod elements separated out
into subrules, but I decided to switch to the single <ws> rule
for the time being since I'm starting to re-use this particular
pattern in a few other tools.
Pm
