Alex Rousskov wrote:
> Shouldn't "index walker" return "read-only" objects? Sounds dangerous to
> provide a write access to objects the order of which is not supposed to
> be changed (from the walker point of view).
>
> The latter would make API different for the two kinds of walkers, but
> that actually may be a good thing! I can hardly imagine a reasonable
> application that uses a "walker" without knowing whether that walker
> modifies the store or not. Making walker "types" different may prevent
> people from doing stupid things like searching for an object with a
> "purge" walker.
Good point.
Hmm.. I'll think about this during the day but I will probbaly make the
proposed change even if it increases the size of the API definition.
From a policy implementation point of view it does not make much of a
difference as the two behaves quite differently anyway.
/Henrik
Received on Tue Apr 25 2000 - 00:17:11 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:24 MST