Re: Metadata (was: Plugin API (was Re: new ideas))

On Thu, 2002-07-11 at 08:37, Andreas Zehender wrote:
> Hi!
> 
> On Thursday 11 July 2002 15:07, Greg Copeland wrote:
> > On the other hand, if only the plugin that creates the meta data is able
> > to access it, one could argue that it's privately owned by the object
> > but this does no one any good and would only serve to waste time in
> > development.  If it's privately owned by the object and accessible by
> > only a single entity, I think we are REALLY saying it's privately owned
> > by the *plugin*.  After all, what value would there be in this when it's
> > all said and done if only the plugin that creates it can access it
> > (sounds more and more like private data to the plugin to me).  Why
> > develop a storage medium and API to support what would be more efficient
> > for each plugin to do on a case by case basis which could then be
> > optimized for the type of data it needed to store/manage?
> 
> But then the plugin has to store a list of all objects and take care, if 
> objects are created or deleted. That's what I don't like.
> 

Actually, I was assuming that there would be some facility for reference
tracking...a per object token or object id number, etc.  Nonetheless,
how does that differ either way.  If it's part of the object, the plugin
will has to track which objects (via direct or indirect methods) it's
working with.  If it's part of the plugin, the plugin simply stores the
data that it needs via whatever method is most appropriate, again,
referring back to the owner object via direct or indirect reference.  In
either case, all things being equal, pretty much the same amount of work
is being done in the plugin.  Furthermore, if the plugin stores it is
self, there wasn't any additional development time to create an API and
storage format which may be less optimal.

From a work perspective, not hardly much difference.

As for the creation/deletion issue, there isn't any way around that. 
There needs to be some form of notification regardless of where the meta
data is stored.  After all, meta data is associated with an object no
matter who "owns" the data.  As such, you can't create meta data for an
object that doesn't exist -- that is, the object should really be
created first.  Furthermore, if an object is deleted, more than likely,
any and every plugin is going to want to know this as it may well have
some form of reference to the deleted object.

> The data should be stored inside the object whether it is protected or not if 
> it belongs to the object.

I do want to stress that I'm simply brain storming here and trying to
better understand where it is you guys are coming from.  As such, I'm
playing devils advocate.  That is, I'm arguing both sides for the sake
of better understand your vision...and helping to validate all of our
assumptions along the way.  Of course, since I keep hearing
semi-conflicted views I'm trying to better understand the reasoning
behind it all.

Greg

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQA9LY/a4lr1bpbcL6kRAgb3AKCER7AzZtyC1dHwSnmDYEymuEXVYgCePMjK
HiRFNHRDkws0pc/RIfVj1xA=
=0dtf
-----END PGP SIGNATURE-----