Relevant facts:
1. bbc.co.uk is the fifth most popular web site in the UK
(http://www.alexa.com/siteinfo/bbc.co.uk)
2. We use DBIx::Class extensively on our iPlayer backend.
3. Performance is critical.
Code in our Web templates is not allowed to touch the database. Thus, for each
DBIx::Class result object, we extract the relevant data and put it into a
simple object. The relevant code looks like this:
sub _init_from_real_thing {
my $self = shift;
# from _real_thing
$self->$_( $self->_dbic->$_ ) for $self->attributes;
return $self;
}
Creating one of these objects takes approximately 50 milliseconds. Each of the
individual channels or episodes you see on http://www.bbc.co.uk/iplayer/
represents one of these blocks. If we create 50 of these objects, that's 2.5
seconds -- way too long.
As a performance hack, I'd like to experiment with something like this:
sub _init_from_real_thing {
my $self = shift;
my @attributes = $self->attributes;
@{$sel...@attributes} = @{$self->_dbic->{_column_data...@attributes};
return $self;
}
Using the slices is a horrible encapsulation violation. I can deal with that
on our side, but not on the dbic side. Is there a clean way of fetching all of
that at once?
If you're curious, here's my benchmark:
Rate methods slice
methods 26954/s -- -71%
slice 94340/s 250% --
Using the slice is almost four times faster than the "clean" strategy.
Cheers,
Ovid
--
Buy the book - http://www.oreilly.com/catalog/perlhks/
Tech blog - http://use.perl.org/~Ovid/journal/
Twitter - http://twitter.com/OvidPerl
Official Perl 6 Wiki - http://www.perlfoundation.org/perl6
_______________________________________________
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]
