Thus spake Chris Withers ([EMAIL PROTECTED]):

> Ty Sarna wrote:
> > call the manage_delete() method on the object. That is, you ask the
> > object to delete itself.
> This seems unnecessarily long-winded to me. What are the reasons for it
> being like this? (examples are good... ;-)
> > However, once you have the object, operations on the object should be
> > done as methods on the object.
> I'm sure it's against Demeter's Law (I've read the Wiki ;-) to work this
> way. You should ask the rack to delete an object and it can process this
> how it wants... most often this would be by asking the object to delete
> itself, but maybe it's be more efficient for the rack to just delete the
> object (for example, if the object is actually a row in an SQL table)

I could be mistaken about this, as I'm still having trouble following the
ZPatterns train of thought, but I think the reason that the object needs to
delete itself is that the object may not be stored in a single place.  For
instance, the object might be created from 5 rows from seperate tables in
an SQL db and 2 attributes gotten from LDAP.  I'm not entirely clear how
asking the object to delete itself in this case is going to improve the
situation over just asking the rack to, but I think that was the intent.


Scott Parish

PGP signature

Reply via email to