RE: [boost] Filesystem attributes

From: Stewart, Robert (stewart_at_[hidden])
Date: 2002-03-05 13:36:30

From: Jan Langer [mailto:jan_at_[hidden]]
> 
> now, we have the focus on the right thing. why does it cost more? your
> approach does create an attribute cache at every iteration cycle.
> independent from the fact whether the user wants an attribute cache to
> be set up or not.
> while my approach needs the explicit creation of an attribute 
> cache. if
> the user just searches a file by name it will be quite easy and cheap.
> if he doesn't create a cache he can be sure what will be done.
> this design will IMO be much clearer.

Obviously, the "attribute cache" could be fetched on demand, obviating the
overhead to which you refer.

Rob
Susquehanna International Group, LLP
http://www.sig.com