On Thu, 6 Nov 2008, Ovid wrote:
Short question (I hope). I have a resultset which inflates objects based upon an id. While
the ID may exist, it's acceptable for the underlying inflated object to not exist.
Unfortunately, this means that the "$rs->count" might report more objects than
actually exist.
This is the correct behavior, but we'd like to be able to create a resultset
with a filter applied which represents what we really have. Something like:
my $new_rs = $rs->filter( sub { defined $_->inflate_entity } );
'$rs->search_rs' doesn't seem appropriate because regrettably, joining across
tables doesn't appear to be an option here.
Cheers,
Ovid
Longer description below for those who are confused.
Note: This is a greatly simplified description of the problem, reducing it to
its core elements.
Customers require all objects to be searchable via a common ID formats. So
let's say we have three tables:
brand
series
episodes
And we have an 'identifier' table (again, greatly simplified):
identifier
----------
identifier_id int primary key
identifier varchar(32)
object_type enum('Brand', 'Series', 'Episode') (yuck)
So if a customer gives me an id, I can check in the identifier table and see
that it's a key for a brand and this requires a second check in the brand table
to fetch the item. Here's the problem: if we delete a brand, series or
episode, we cannot delete the identifier because it must not be reused. Though
some of these are custom methods, you should be able to understand the problem
from here:
my $entity = $schema->resultset('Identifier')->search(
{ type => 'pid' }
)->first->inflate_entity;
my $pid = $entity->pid;
$entity->delete;
my $rs = $schema->resultset('Identifier')->search({
type => 'pid',
value => $pid,
});
print $rs->count; # prints 1, even though the $entity is gone
This part of the description makes it sound like your constraints/filter
criteria are all SQL-based.. That essentially you need a more complex
search that ends up similar to (psuedo sql):
select * from Identifier
where type='pid' and value=$pid
and (exists select pid from brand
or exists select pid from series
or exists select pid from episodes)
- which will only return used ones.
As for doing this on an rs where the condition is NOT database based, all
I can think of is: add cache => 1 to the attrs, force the rs to fill the
cache by calling ->all on it or something, then post-manipulate the cache
contents on the $rs object.
->count doesnt use the cache that I know of, tho one could override it to.
other methods should, and if you used prefetch as well it should happily
assume your manipulated cache is correct.
see set_cache/get_cache on ResultSet.
btw your inflate_entity sounds a lot like inflate_result, see
::Manual::Cookbook for an example.
Jess
_______________________________________________
List: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/dbix-class
IRC: irc.perl.org#dbix-class
SVN: http://dev.catalyst.perl.org/repos/bast/DBIx-Class/
Searchable Archive: http://www.grokbase.com/group/[email protected]
