[dev] thoughts on the prefs problem

[Brent J. Nordquist]

On Tue, 20 Feb 2001, Jon Parise <jon@csh.rit.edu> wrote:

> On Wed, Jan 10, 2001 at 06:20:35PM -0500, Jon Parise wrote:
>
> > > and thus lets you have prefs that only last the length
> > > of the session?
> >
> > That's a good idea.  Ideally, however, all of the preference drivers
> > should use session storage for their data.
>
> I've basically closed this issue with the introduction of the
> preferences "cache".  The preferences subsystem now behaves optimally
> in the case of both reading and writing (meaning it only hits the
> preferences storage server when there is know new or stale data).

Very cool!  A transparent cache sounds like a very nice performance
optimizer.

However, preference caching is really orthogonal to preference
persistence, isn't it?  Whoever made the above comment was asking for
preferences that last only the duration of one session (like a temporary
override); is there a way to mark preferences such that they aren't kept
when the session ends?

(I don't think I care whether that particular feature is implemented; just
wanted to point out that it's different from caching.)

-- 
Brent J. Nordquist <bjn@horde.org>
Yahoo!: Brent_Nordquist / AIM: BrentJNordquist / ICQ: 76158942