On Thu, 6 Mar 2003, fglock wrote:
> This is an idea to make extensible iterators.
>
> SYNOPSIS
>
> $iter_user = DateTime::Iterator (
> truncate => sub { ..... },
> next => sub { ..... },
> previous => sub { ..... },
> );
I take it these subs would do something like receive the current datetime
object and then return a new one?
> $dt->next( $iter_user );
> $dt->previous( $iter_user );
> $dt->truncate( $iter_user );
I'm not sure I like the idea of creating an iterator object and then not
actually using it directly. I also don't particularly like to the idea of
having one iterator contain multiple code references. It should just do
one thing.
What about ...
my $iter = DateTime::Iterator->new( datetime => $dt,
iterator => sub { ... },
);
# next does not mean "a later datetime" here, just the next in the
# iteration.
my $next_dt = $iter->next;
For many uses, this would be sufficient:
my $iter = DateTime::Iterator->new( datetime => $dt,
duration =>
DateTime::Duration->new( days => 1 ),
);
> $dt->next( 'day' );
> $dt->previous( 'day' );
> $dt->truncate( 'day' );
I definitely don't think of truncation as related to iteration. The two
have basically nothing in common.
Back when I started this project, I was thinking that iterators would be
produced by sets, so that we might have this:
my $set = DateTime::Set->new( recurrence => sub { ... },
start => $dt );
my $iter = $set->iterator;
while ( my $dt = $iter->next ) { ... }
I still like this API. What do others think?
-dave
/*=======================
House Absolute Consulting
www.houseabsolute.com
=======================*/
