Hi,
ok, so from your agreement to many points I wrote about desktop:// and
semdesk:// URIs,
I read that you like the idea :-)
now.... to get this running we need to make the two RFC's stable....
because if we go this way, the documents must be understood and
supported by you also (=we build consensus around the documents)
where can we host the documents further? nepomuk-svn is fine, any other
ideas (is there a place on freedesktop.org where we could further work
on this documentation about URIs?) I have no time to find out, but I
have time to help the editing process of standard documents (I did this
before for w3c).
The discussion we have here MUST go into the documents!
(one answer below)
It was Hari krishna Anandhan who said at the right time 30.09.2008 19:11
the following words:
On Mon, Sep 22, 2008 at 9:35 PM, Leo Sauermann <[EMAIL PROTECTED]> wrote:
The desktop uris are addressed here:
http://aperture.wiki.sourceforge.net/SemdeskUris
The semdesk-uris here:
http://dev.nepomuk.semanticdesktop.org/repos/trunk/doc/2008_09_semdeskurischeme/index.html
The second approach shows how we wanted to have a global identification
scheme, retrievable,
secure, and ready to cope with multiple devices and users.
Its based on a trick: we use XMPP underneath.
This is probably the best way to do it, but its not clearly defined yet,
I think it would be better if we can use both XMPP and the aperture
(Desktop uris) for generating URIs.
As XMPP requires an explicit server, it makes more sense in an
environment where user ids are more controlled (like in corporates or
schools), but still for desktops, I think we need something like
aperture (with user, host and maybe application ids in URI).
And we need to identify how to handle the scenario where there are
clashes in the URI parts when things are transfered to another machine
or user...
On Tue, Sep 23, 2008 at 12:40 PM, Leo Sauermann <[EMAIL PROTECTED]> wrote:
for files, we should stick to the iana-registered file:// uri scheme
for the information elements inside files we may use hash-based identifiers
or embedded uuids in files (such as email message-ids)
for elements stored in desktop applications already, and accessible via
dbus, and never shared or sent with other computers or people, we may come
up with something like kde:// or desktop://
makes perfect sense ;)
for things defined inside the user's personal information model (= tags,
people, topics) which can be shared across computers and people, the
"nepomuk URI scheme" is needed,
exactly.
ok, for nepomuk-uris (or "thing"-URIs):
* a once-auto-generated 128 bit key (either UUID with MAC, or GPG) will not
help to identify the user, machine, etc. or dereference the URI for
retrieval in the future
what would be the possible usecase where dereferencing the URI might
make sense ?
Some examples:
To update information shared with someone
"Bob wants to know if the URI's content changed on Alice's machine(s)"
to get Binary files from others
"Bob like 'Julia Nune's Buttercup' bootleg that is stored on Alices
computer and shared with bob. he makes a 'give me the binary attached to
this semdesk uri' call and Alice's machine starts streaming the mp3 to Bob"
Therefore I proposed the "semdesk" uri scheme, as described here:
http://dev.nepomuk.semanticdesktop.org/repos/trunk/doc/2008_09_semdeskurischeme/index.html
which:
* is globally unique because of the underlying [EMAIL PROTECTED] scheme taken
from
XMPP/Jabber (thus covers the UUID/MAC approaches)
* allows using multiple devices for one user (also technically for
dereferenciation, as XMPP allows multiple agents connect using the same
identifier but different "profiles")
* but is human readable ([EMAIL PROTECTED], thats easy to read and to debug if
something goes wrong)
Perfectly valid. But, this alone might not work in scenarios where the
control is not handled by a server (like in jabber case).
for this you suggested a "local" jabber-server approach,
something like "[EMAIL PROTECTED]" or "[EMAIL PROTECTED]"
this is ok
We may also
have to think of scenarios where adhoc groups of people transfer files
from one pc to another (like why can't I transfer my photo albums and
related metadata to my sister's pc. here there may not be an internet
connection (and hence a jabber server) available at all)
The sentence starts with "we may also have to think" and contains many
mays and possibilities.
The case described is managed by protocols such as rendezvous,
and you can't make one size fit all:
uris that identify a person's information model elements are great to
exchange information
in normal settings (e-mails, via web2.0 platforms), and the rendezvous
case we can safely ignore.
It will not be easy to find a uri scheme that works both globally,
on one desktop, and in p2p/rendezvous networks, because their inherent
structures
are different.
so, for the sake of moving on: I would cut that feature OUT.
best
Leo
Cheers,
Hari
_______________________________________________
nepomuk-kde mailing list
[email protected]
http://lists.semanticdesktop.org/mailman/listinfo/nepomuk-kde
--
____________________________________________________
DI Leo Sauermann http://www.dfki.de/~sauermann
Deutsches Forschungszentrum fuer
Kuenstliche Intelligenz DFKI GmbH
Trippstadter Strasse 122
P.O. Box 2080 Fon: +49 631 20575-116
D-67663 Kaiserslautern Fax: +49 631 20575-102
Germany Mail: [EMAIL PROTECTED]
Geschaeftsfuehrung:
Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff
Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes
Amtsgericht Kaiserslautern, HRB 2313
____________________________________________________
_______________________________________________
nepomuk-kde mailing list
[email protected]
http://lists.semanticdesktop.org/mailman/listinfo/nepomuk-kde
