Saying "sorry to be late" is lame after a month of non-activity on this
topic,
I just can't invest much time into this topic, and I would like to
contribute more,
but I need a good process for that, see more below on a suggestion how
we can do this a bit better.
Whats the status of this today?
Did you implement something already?
I understand Sebastian's argument of "we have to be pragmatic",
But until I know all requirements ("it works offline from minute one") I
did assume some freedom on requirements...
I still think we can't solve all problems at once: single desktop, p2p,
central server AND being retrievable (uris should be retrievable, not
just random numbers)
I am referring now to the draft spec I edit:
http://dev.nepomuk.semanticdesktop.org/repos/trunk/doc/2008_09_semdeskurischeme/index.html
It was Sebastian Trüg who said at the right time 02.10.2008 10:58 the
following words:
In a world where cloud computing and mesh services (where data is
shared between multiple devices of the same person) are gaining
ground, i think we have to think of how to solve all the three
scenarios
1. single desktop
2. p2p
3. central server
I understand that a single URI might not be possibly used in all the
scenarios. But, why not have multiple URIs for the above scenarios for
the same object?
We can retain the current URI scheme for the single desktop scenario.
Then, we can use an additional GloballySharedURI for resources shared
via a central server. And, maybe a PeerSharedURI for p2p scenarios.
all the above URIs can be mapped to a single resource as needed and
the corresponding URI could be used for the corresp scenario.
technical question: how do we implement this? I mean, not everybody has a
jabber id and asking all users to create a jabber id before they can start
tagging their stuff seems overkill. So how is this handled? Start out with
plain URIs like nepomuk:/foobar or with local ones like
nepomuk:/[EMAIL PROTECTED]/foobar? And what to do once we have a jabber id and
want to exchange? Update all the URIs?
There can't be a solution to this, we can never cover all aspects in one
scheme.
We are still talking about URIs for tags and "Things" defined in the
user's Personal Information Model (PIMO),
(cancelling out public ontology uris and cancelling out file uris)
Looking back we wanted all of these features at once:
* identifying the Thing on one machine and being able to retrieve/open
on this machine - dereferencation and creation similar to file:// URIs
* identifying the Thing on a global network with a jabber-server or
HTTP-server like infrastructure - dereferencation and creation similar
to http:// URIs or jabber ids
* identifying the Thing on a p2p network based on a hash-sum or a
[EMAIL PROTECTED] identifier - dereferencation and creation similar to magnet
The last URI schemes for p2p are described in the cool uris document here:
http://www.w3.org/TR/cooluris/#otherideas
file and global identifiers are following [EMAIL PROTECTED] schemes, the p2p
things follow hash-ideas.
But we can't really use the p2p uris, because they assume the use of
Hashes, and hashes only work good on static, non-changing, big
multimedia files (=mp3s, moviez, warez).
An identifier based on a hash for a thing would be awkward because its
hard to define on what to hash, and hashing RDF graphs is a tricky
thing, and the graph will change.
leaves file and http-uri styles. From these standards we can see that
both work also when the "server" part is removed.
Rethinking this, I would NOT suggest to replace the <hostname> variable
in semdesk uris with "localhost", because that seems to say "localhost
is a correct hostname which is tricky - rather change the syntax of the
URI sot hat GREPs and regex will have to stick to it.
I changed the recommendation to be able to include something like this:
semdesk://bob@/things/alice.jpg
please comment.
see
http://dev.nepomuk.semanticdesktop.org/repos/trunk/doc/2008_09_semdeskurischeme/index.html
And Leo, please don't say again that we cut this issue since it is important.
oh, sorry, misunderstanding, I meant to cut the p2p feature!
and maybe the social feature!
but not the "URI" feature.
We are in the real world here, not in an academic prototype.
hey, you are right.... but it hit a nerve in my past:
/begin sarcasm and nitwitty remarks
I never thought nepomuk as an academic prototype,
First I didn't think about academic funding,
but there was no venture capital funding in 2003 to make this happen.
So I had to go for academic funding money, which turned out to be a good
environment, it gave us the chance to start something really ahead...
and work with some of the best brains in the field.
Of course, these brains are weird, and,
OMG, I changed into a scientist....
repent, repent!
/end of sarcasm and nitwitty remarks
8-]
well,
this is standardization work, it just takes time
and a well-structured process.
If nobody of you volunteers for it, I could play the editor for the
"desktop and semdesk uri scheme" documents,
I would then use the proven W3C way of standardizing stuff:
request for comments until happy.
I would put the documents in a status of "editor's draft"
and then would ask you for formal "comments"
which I then edit back to the document.
After a second round, we would be somewhere.
It will take around two months at minimum to get feedback and work it in.
but I can't work much quicker based on my various other responsibilities
and my job.
Do you want me as editor for the specs?
If yes, say so and I will announce the deadlines for feedback to you and
others.
or do it somehow else, I see e-mails are a somehow slower process...
because no one is stuck to deadlines.
best
Leo
Users will get
angry and spam the help and bug forums if we force them to create a jabber
account on first login. We need to start from a disconnected solution. Sad
but true.
Cheers,
Sebastian
_______________________________________________
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
