On Fri, Jul 09, 2004 at 12:45:05AM -0400, Bill Clark wrote:
On Fri, 9 Jul 2004 06:10:46 +0200, Tomasz Wegrzanowski
<taw(a)users.sf.net> wrote:
Coding metadata as categories works just as well.
It may work just as well for the concept of 'category' but it doesn't
work as well as a custom tool would, from an implementation
standpoint.
There are enough stubs that we're in no danger of running out. The
real problem is that there's no convenient interface for searching
through all of them -- EXCEPT by using the stub category interface.
But the same list can simply be generated from a backup copy of the
database, and formatted so that it has the same type of format as a
category view would. It wouldn't be real-time, but simple perl
script could create the proper files from periodic database updates
(and that should be more than sufficient for our purposes).
Stubs are bad enough without having them chew up cycles on the central
database too.
Why should we use such a complicated setup (with perl scripts and backup copies)
when simple category would work just as well ?