Re: [GNUnet-developers] namespace

[Glenn McGrath]

On Wed, 17 Jul 2002 23:59:23 -0500
Christian Grothoff <address@hidden> wrote:

> How about storing a date/timestamp in the index file?
> 

Yes, but a search can only compare results that were found, there may be
newer index file released that the search didnt find.
Having a namespace, an "offical" location for a document means you know
that whatever is there is the most current.


> If you put the host id in the path, it is pretty clear from whom the
> content is coming from (host-id is in some sense equivalent to an IP).
> OTHO, publishing an urn does not imply that the content came from that
> entitiy, just that that entity is aware of the content. Still this makes
> you a slightly more visible target for a legal attack (which may be ok
> in the software distribution scenario).
> Also how would you enforce that other hosts can not publish anything in
> this namespace? 
> 

hmm, i thought host-id was the has hash of the public key, similar to a
gpg fingerprint, myIdentity in keyservice.c, that must be something else.
I guess a host id has to map back to an ip somehow. 

If the namespace-id was based on the public key then anything in that
namespace can be signed by the public key owner so it can be shown to be
legit.


> 
> Is a urn bound to the host that it resides on or does it move in the
> network? If urns move, how do you do the distributed update? Why do you
> want the (useless) "urn:gnunet" header? gnunet:// should do if we really
> need a prefix (stick to common notations!). How would the urn differ
> from a web-site (http) containing the references?
> 
> If urns do not move, what happens if the host goes offline? Does all the
> 
> content it linked to become unavailable? 
> 

I dont really care how a namespace is formated, i was just following what
i thought was standard.

I think it would be better if the urn was bound to a user rather than a
host.

> How do users find the "path" part of the urn? How do users get the urn
> in the first place?
> 

It wouldnt matter how the urn was distributed, its purpose is just to be a
stable point of reference.

I dont know exactly how to achieve what i describe, just throwing the idea
around, maybe it could be a distributed hashtable that maps urn to gnunet
hash values.

Do users have public keys, or just hosts ?


Glenn

pgpYaUz2GGiwI.pgp
Description: PGP signature