Re: Private Class Methods

Method calls bypass Exporter, so no, this won't work.

There are several ways to do "private" methods.

The standard way is to not bother enforcing private-ness. Just pre-pend an underscore (_) to your method names and let the user consciously make the choice to violate encapsulation.
You can put a
```
sub _my_private_method {
    my $self = shift;

    die "Private method, jerk!"
        unless UNIVERSAL::isa($self, __PACKAGE__);

# ... Continue on here as normal
}
[download]
```
That's known as a gatekeeper. I haven't seen that a lot. That kind of private allows inheritance to work. The other way, requiring that you must be of that class and that class alone would be to do something like ref $self eq __PACKAGE__, more akin to C++'s private.
This applies more to data, but you can change your reference from a hashref to a coderef, effectively making all your object instances closures. This would make all your data private. You could extend it to methods, I suppose. I wouldn't reccomend it because it's extremely non-standard and would be very hard to maintain. There's also a minor speed hit, but it's not that major.

All in all, the best method I've found is #1. Just trust your users. It's the easiest policy to implement.

------
We are the carpenters and bricklayers of the Information Age.

Don't go borrowing trouble. For programmers, this means Worry only about what you need to implement.

Comment on Re: Private Class Methods Select or Download Code

Replies are listed 'Best First'.
Re: Re: Private Class Methods by perrin (Chancellor) on Apr 05, 2002 at 21:43 UTC
++ on #1. Don't waste your time on things like this. Name it _foo() and be done with it.	[reply]
Re: Re: Private Class Methods by RMGir (Prior) on Apr 06, 2002 at 14:27 UTC
++ on #1. But #2 will fail on "indirect" private calls, i.e. where class A implements `sub public_method { # do something $self->_my_private_method(); # do something else }` [download] and then public_method is invoked on a derived object. Bad bad. -- Mike	[reply] [d/l]