On Wed, Mar 15, 2006 at 06:50:53AM +0800, Audrey Tang wrote:
> Heya. Today I noted with delight that PGE supports interpolation of
> closures:
>
> $ parrot demo.pir
> rule Foo {{ print "Hello" }}
>
> Though the Perl 6 form is {...} instead of {{...}} -- is that intentional?
Yes, it's intentional. PGE uses a generalized form of {{...}}
for generic closures in any target language, as mentioned by
Larry in a post from last summer. We're reserving { ... }
for the specific case of Perl 6 closures (where the embedded
code is parsed as Perl 6 code), but PGE allows any target
language to be embedded in a rule by using {{ ... }}.
Or, if the embedded closure itself needs to double braces
for some reason, then the rule can use {{{...}}}, {{{{...}}}}, etc.,
similar to how POD handles <<<...>>>.
So, one can have
rule myrule :lang(PIR) { <token> {{
I0 = match
I0 *= 2
print I0
}} }
and the embedded PIR "closure" will be executed after the
<token> subrule is matched. The :lang() modifier can have
any compiler that Parrot's "compreg" instruction understands.
> We have started self-hosting Perl6 compilation, beginning with a
> self-hosting Rule engine; to construct the AST object, we use the
> "return" form recently specified in S05.pod. Please see
> http://perlcabal.org/~gaal/peek/slide37.html for an illustration of its use.
I haven't had a chance to fully look at the change to S05, so
give me a day to look it over and I can respond more intelligently
about how quickly we can get PGE to support the changes.
> Is it possible for PGE, with its embedded code closure, support
> something like that? For example, make a lexical symbol RETURN
> visible inside the interpolated block, which will set the result object
> and return a match success at that position?
>
> rule Foo {{
> $P0 = new .Some::Tree
> RETURN($P0)
> }}
But in answer to your question, for :lang(PIR) PGE already
puts a nice pre-amble around the PIR code and makes the current
match state available as a "match" symbol, so it seems like
it ought to be possible to do what you're describing here.
> Pointers to where in the PGE source to hack this together would be
> appreciated. :-)
Take a look at the closure.t tests in t/compilers/pge/p6rules,
and at the PGE::Exp::Closure object at the bottom of
compilers/pge/PGE/P6Rule.pir to see how PGE is currently
handling it.
Also, I should be able to start hanging out on #perl6
on a regular basis again, so you can ask questions of me then.
I'll make a point to be available tomorrow morning
(probably around 1500 UTC).
Thanks!
Pm
