On Thursday, November 21, 2002, at 11:22 AM, Tanton Gibbs wrote:
Or, using hypothetical vars:It would be very useful in the second half of a s///, too (grab a match, rewrite it in a canonical form):$s ~~ s/<number>/<number:MoneyFormat>/;once again, I think it could just be $s ~~ s/<number>/$1.as(MoneyFormat)/
$s ~~ s/ $num := <number> /$num.as(MoneyFormat)/;
(This is a bit off-topic, but it's a tiny bit sad that you can name subrules, but you can't refer to them by name in the interpolation unless you use hypothetical vars. Eh, no biggie: it would introduce too many problems to allow it.)
Still, the idea of having "input rules" and "output rules" that are conceptual mirrors of each other just seems so... sweeeeet. I find the syntax
$s ~~ s/<number>/<number:MoneyFormat>/;
or
$s ~~ s/<number:$n>/<$n:MoneyFormat>/;
or even
$s ~~ s/<number>/<$1:%-4.2f>/
to be wonderfully appealing.
On Thursday, November 21, 2002, at 11:49 AM, Dave Whipp wrote:
I thought so too, but I haven't been able to find it again. If anyone has a link, please let me know.The inverse-regex was discussed on perl6-language a few months back (I think the context was an alternative to pack/unpack): I don't recall any definitive conclusions. A simple proposal to pass a match object/hash as an input to the rule falls foul of too many reverse ambiguities. But a constrained regex syntax could be made to work.
MikeL
