Bug #112659 for Memoize-Memcached: Two memcached round-trips for each cache hit

> Memoize.pm has this: > > if (exists $cache->{$argstr}) { > return @{$cache->{$argstr}}; > } else { > .. > } > > So first exists is tested to see if we have a cache hit, and then the value > is fetched. > > Memoize/Memcached.pm has this: > > sub FETCH { > my $self = shift; > my $key = $self->_get_key(shift); > return $self->{memcached_obj}->get($key); > } > > sub EXISTS { > my $self = shift; > return defined $self->FETCH(@_); > } > > So if there is a cache hit, first EXISTS is called (which calls FETCH > internally) and then afterwards, FETCH is called again. Here is what it > looks like in a NYTProf flamegraph: > > [image: Inline image 1] > > ( http://i.imgur.com/WYcDbiK.png ) > > for: > > sub foo { > return "2"; > } > > memoize_memcached('foo'); > foreach (1..100000) { > foo(); > } > > I don't know the side effects, but it looks like execution time for a cache > hit (the common case?) could be cut in half rather easily... > > Peter