On Tue, 2004-02-24 at 11:02, John Beppu wrote:
> The has() method in this release has a bug.
>
> Here's how it looks right now:
>
> sub has
> {
> my ($class, $define) = @_;
> scalar grep { exists $$sdl_config{$_}{$define} } keys %$sdl_config;
> }
>
> Unfortunately, even if certain attributes are set to 0,
> the existence check still passes. For example, I don't
> have SDL_ttf on my system, and the $sdl_config hashref
> knows this. However,
>
> SDL::Config->has('SDL_ttf')
>
> returns 1;
>
> Get rid of "exists" and it'll work as advertised.
True, but it'll eat up memory in the hash if you pass undefined
subsystem names. That's why the original version was:
sub has
{
my ($class, $define) = @_;
while (my ($name, $subsystems) = each %$sdl_config)
{
next unless exists $subsystems->{ $define };
return $subsystems->{ $define } ? 1 : 0;
}
return;
}
Maybe it was silly to optimize this never to use any more memory or to
keep an O(n) operation O(n/2) or better, but it's not susceptible to
this bug.
On the other hand, I suspect the new version performs better in list
context.
-- c
