Re^5: Inheritance - calling parent constructor

But you only know whether something is wrong if you actually peek in the code of the parent class¹. But that means breaking encapsulation², which you aren't supposed to do when doing OO.

I would leave in the bless. At worst, it's redundant. At best, it fixes something the parent does wrong.

¹ And keep peeking. If it's ok now, it may no longer be ok tomorrow after the sysadmin upgraded some packages.
² Of course, the code of the OP already assumes the parents constructor returns a hashref. So the OP has already sacrificed one the pearls of OO already.

Comment on Re^5: Inheritance - calling parent constructor

Replies are listed 'Best First'.
Re^6: Inheritance - calling parent constructor by ikegami (Patriarch) on Dec 30, 2009 at 00:51 UTC
But you only know whether something is wrong if you actually peek in the code of the parent class No, no peeking necessary. I have no idea what you're talking about. At worst, it's redundant. It's usually redundant, but it's not the worst case. At worse, the parent (and thus the child) malfunctions (obviously or subtly). At best, it fixes something the parent does wrong. At best, it's a workaround for a bug in the parent? If there's a bug in the parent, fix the parent. Don't use a workaround just in case you might need it.	[reply]
Re^7: Inheritance - calling parent constructor by JavaFan (Canon) on Dec 30, 2009 at 03:00 UTC
If there's a bug in the parent, fix the parent. That assumes you control the parent.	[reply]
Re^8: Inheritance - calling parent constructor by ikegami (Patriarch) on Dec 30, 2009 at 12:48 UTC
I know, which is why I made an allowance for when the assumption is wrong. ("Don't use a workaround just in case you might need it." allows for using a workaround when needed.) You, on the other hand, assume that all base classes are buggy and permit no exception (because it would break encapsulation!?)	[reply]