RE: Icons of program

[Andreas Kostyrka <andreas ag or at>]

> > Toshio Kuratomi <badger@prtr-13.ucsc.edu> writes:
> > Also -- by defining a very specific icon file format, we are tying
> > ourselves
> > to a specific set of features.  Maybe we put everything we think we
> > could
> > possibly want into the image file -- but ten years from now (When
> > everyone is
> > actually using GNOME :-) we may find that there's something new that
> > we want
> > -- Maybe there's a new way to do 3D modeling that's small enough to
> > make
> > sensible icons or a new compression algorithm that is just world's
> > better --
> > and if our "icons" just contain pointers to the real image, (that is
> > then
> > decodable by some general image library like imlib..)  then all we
> > have to do
> > is start defining pointers to the new image type.  If we put the image
> > data
> > into the icon file along with the rest of the meta-info, we have to
> > scrap the
> > icon format and define version-2 (and 3 and 4....)
You still have to define version-2 of the format so the new pointers are
recognized. And just HOW do you copy such information (and automatically
reconcile it) to removable media?

> > Just view it (sort of) as a database problem, an normalise the
> > file(table).  The images shouldn't be in there for the reasons that
> > you don't build your database as one big table.
But it's not a database problem. And if you have ever done db programming,
than you know that sometimes violations of ER normalization are needed.

Andreas