Again, I apparently gave the impression that all document files would
simply have their icon information supplied as part of some "header" to the
file. Such is not the case in the scheme described.
The document simply provides the system with a four byte identifier (well,
actually eight bytes if you include the creator information) which the
system uses to determine which icon to use. No, there are no "set" icons,
the system gets the icon information from the creating application–which
keeps one (and only one) copy of the icon. Disk space is conserved, no
info files appear to clutter up the CLI display, and the only additional
overhead is eight bytes of (usually invisible) file identifier information
for each file–certainly much better than the current 400 plus bytes taken
up byte the .info files.
James,
Ahh okay I understand you. Still, the trouble is with backward compatability.
But a future machine could/should certainly use a scheme like that.
On the other hand, hmmm, if a special identifier code was used to … no never
mind, still not compatable. Shoot. Hindsight is great <grin>.
Kev