Tired of Hierarchies
Aryeh M. Friedman
aryeh.friedman at gmail.com
Mon May 5 02:20:35 UTC 2008
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Jason C. Wells wrote:
| Aryeh M. Friedman wrote:
|
|> ~ 5. If the system didn't plan for some major catagory it will be
crunched into a sub catagory(s) that do not make very much sense for
example under Library of Congress computer science is under math
(QA76.XXXX) but electronics is under TK510[456].XXXX
|
| In this example the call number performs a dual function of identifier
and grouping. The ability to lookup the address in a computerized card
catalog database mostly negates the weakness of the poor grouping.
Because a computer can manage location and grouping in some other
fashion, all we really need is a unique identifier.
|
|> A very good example all the items above is the current ports
system. In short the more finally cut we make our categories the
harder it is guess/generate the "search key" (either a real key or
metaphorically a mental picture of one). For all the above reasons I
would argue for flatter hieracies with metahierachies overlayed for
different purposes then one typically sees today.
|
| The idea of an overlay I think is a very powerful one. The file
system hierarchy could simply be one overlay that might be applied by a
hypothetical storage manager. An author might use an author's overlay
suitable to the author's task. All user's would have to be careful to
divorce that idea of "what" they are looking at from "where" they found
it. There would multiple disjoint locations in an overlay system that
all refer to precisely the same resource.
|
One issue you would need to consider in such a system is how to ensure
that the "address" in one hierachy doesn't interfere with the address in
an other hierachyy (the two are allowed to vary independantly). Two
examples pop to mind immediatly URL's and Email addresses.
Specifically both refer to a resource that theortically has nothing to
do with it's physical location but only on it's conceptual location.
For example I should give a damn what machine someone uses to receive
mail on in order to send them mail, namely how many times I switch ISP's
my address would stay "aryeh.m.friedman" (for example) [notice no
@dns]. Back in the late 90's I offered a service like this
http://web.archive.org/web/*/http://www.atless.net but sadly not enough
demand to keep it going. The URL issue is the same.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.9 (FreeBSD)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEARECAAYFAkgebuYACgkQk8GFzCrQm4CgFQCfUBDce8iqlJnGcRban3OIaWLW
l5AAoNI9KVWNSnQ+9DJA/aKd/zPGyNar
=MnXY
-----END PGP SIGNATURE-----
More information about the freebsd-chat
mailing list