[Opensim-dev] Global identifiers
diva at metaverseink.com
diva at metaverseink.com
Sun Aug 29 16:55:19 UTC 2010
What Melanie is saying is that the URI stored in persistent storage is
both the global reference (the part http://authority/user/uuid) and a
simple cache (the part /Melanie+Milland).
This doesn't prevent grids from looking up the global reference part
(i.e. update the cache) whenever they wish. And it does, indeed, reduce
lookups significantly in the context of OpenSimulator.
I like it.
Mike Dickson wrote:
> Personally I think attaching additional semantics beyond just a unique
> ID that is associated with an Agent Service is a bad idea. Beyond that
> a simple rest service definition that allows quick lookup and caching of
> id to user friendly names would solve the problem and in that case the
> service can make a decision on behalf of the user as to what information
> is valid to be shared (assuming a trust relationship exists betweem the
> grid services trying to access the information.
>
> So IMO an agent id is simply a URL that identifies the scope (the agent
> service) and id of a user. The service can then export whatever service
> interfaces are appropriate to allow access to additional data.
>
> Mike
>
> On Sun, 2010-08-29 at 16:13 +0000, Ideia Boa wrote:
>> We see the same problem than Zonja with our grid
>>
>> On 29-08-2010 5:27, Zonja Capalini wrote:
>>> I see one small problem with this approach: UUIDs are immutable,
>>> but it's conceivable that a world operator could allow certain form
>>> of
>>> updating of user names, while still retaining the same identity
>>> (I've had to manually edit user names in some cases in the worlds
>>> I administer, for a number of reasons).
>>>
>>>
>>> In this scenario, if an URI is resolved to a name that has changed
>>> this can potentially require a lot of updates in the database
>>> (e.g., if the foreign user has created many objects in the local
>>> world).
>>>
>>>
>>> OTOH, if the URI -> username association is stored in a different
>>> table,
>>> this table can also keep other, valuable, information, for example
>>> the
>>> date of the latest resolution, whether the world appears to be
>>> active atm, etc.
>>>
>>>
>>> /Zonja
>>>
>>> On Sun, Aug 29, 2010 at 1:44 PM, Melanie <melanie at t-data.com> wrote:
>>> We should.
>>>
>>> Also, we should use extra info in the URI. Reson:
>>>
>>> http://www.avination.net:8004/user/44626b40-13d6-4817-b61b-de5df7b5e7e8
>>>
>>> The above is totally meaningless. It can't be used to do
>>> anything
>>> with unless www.avination.net exists and points to a
>>> gatekeeper.
>>>
>>> However,
>>>
>>> http://www.avination.net/user/44626b40-13d6-4817-b61b-de5df7b5e7e8/Melanie+Milland
>>>
>>> makes more sense here.
>>>
>>> The URI itself provides a "Display name" that the resolver
>>> at that
>>> URL can treat as extra path info and ignore, if it chooses.
>>>
>>> This would allow us to create a temporary memory cache
>>> record of the
>>> UUID -> name mapping that would let us display a prim
>>> creator
>>> without a lookup, which is a potentially frequent process.
>>>
>>> The sim can take the URL at face value and diassemble it,
>>> using
>>> 44626b40-13d6-4817-b61b-de5df7b5e7e8 -> "Melanie
>>> Milland at www.avination.net" for the cache and returning that
>>> to the
>>> viewer as the creator, all without a lookup.
>>>
>>> While this doesn't prevent verification of stale URI's from
>>> failing,
>>> it does allow to display a meaningful text if that happens.
>>>
>>> Melanie
>>>
>>> diva at metaverseink.com wrote:
>>> > Looks like ppl are reading more into this discussion than
>>> I intended.
>>> >
>>> > The hypergrid is up & running with all authentication and
>>> security in
>>> > place, and so are exchanges of content via HG and
>>> archives. What's
>>> > missing is *systematic* global identification of
>>> resources. OpenSim
>>> > already does that internally for resolving *certain*
>>> identifiers on the
>>> > Hypergrid, but nothing is stored persistently yet. That is
>>> going to
>>> > change soon, because 1) I want to make friends & IM work
>>> across the HG
>>> > (so, for example, your foreign friend needs to be
>>> identified by a global
>>> > ID); and 2) we really need to fix the b0rked "creator"
>>> field in OARs/IARs.
>>> >
>>> > This means that we need to write URIs persistently, both
>>> in certain
>>> > fields of the DB (which is already prepared for what's
>>> coming) and in
>>> > the archives.
>>> >
>>> > So the issue here is really narrow. Assuming everyone
>>> agrees that we
>>> > should use URIs, should we add type information in the URI
>>> or not? Any
>>> > other thoughts on the *form* of these URIs?
>>> >
>>> >
>>> > mysticaldemina at xrgrid.com wrote:
>>> >> May be good to share what your use case is. As universal
>>> are you suggesting
>>> >> an identifier that separate, potentially un-trusted
>>> domains, would use to
>>> >> identify the same person?
>>> >>
>>> >> Is so I don't think you can do that with two parties, you
>>> need at least one
>>> >> more party to validate that they are the same person,
>>> like how we do with
>>> >> SSL certificates, or with some kind of authentication,
>>> like you send me an
>>> >> email address which gets me to a profile, but I still
>>> need to enter in a
>>> >> password or something to get access to that profile.
>>> >>
>>> >> M.
>>> >>
>>> >> -----Original Message-----
>>> >> From: opensim-dev-bounces at lists.berlios.de
>>> >> [mailto:opensim-dev-bounces at lists.berlios.de] On Behalf
>>> Of Ai Austin
>>> >> Sent: Saturday, August 28, 2010 4:59 PM
>>> >> To: opensim-dev at lists.berlios.de
>>> >> Subject: Re: [Opensim-dev] Global identifiers
>>> >>
>>> >> diva wrote:
>>> >>> I'm about to introduce global identifiers, so that I can
>>> make friends
>>> >>> and IM work on the hypergrid, and would like feedback on
>>> the best form
>>> >>> of these identifiers.
>>> >>
>>> >>> Here are some options:
>>> >>> ... Thoughts?
>>> >>
>>> >>
>>> >> A couple of thoughts and observations Diva...
>>> >>
>>> >> Could the taxonomy of "types" you use cause problems if
>>> the chosen
>>> >> 1-1 mapping for a UUID is not felt to work well i future.
>>> >>
>>> >> "user" is also perhaps a different notion to a specific
>>> "avatar"
>>> >>
>>> >> It would be nice if any UUID in a URI you use can be
>>> resolved (e.g.
>>> >> to the avatar name) by any host that has the mapping
>>> (like the
>>> >> distributed nature of DNS works), so its not dependent on
>>> the host
>>> >> continuing to exist, or to be up at the time information
>>> on the
>>> >> avatar is sought.
>>> >>
>>> >> AS an example, we have shifted our data bases between
>>> machines and
>>> >> have done so 3 or 4 times since we started running
>>> OpenSim, carrying
>>> >> the UUIDs of avatars (and the UIIDs of regions we use)
>>> forwards to
>>> >> the new data bases.
>>> >>
>>> >> _______________________________________________
>>> >> Opensim-dev mailing list
>>> >> Opensim-dev at lists.berlios.de
>>> >> https://lists.berlios.de/mailman/listinfo/opensim-dev
>>> >>
>>> >> _______________________________________________
>>> >> Opensim-dev mailing list
>>> >> Opensim-dev at lists.berlios.de
>>> >> https://lists.berlios.de/mailman/listinfo/opensim-dev
>>> >>
>>> > _______________________________________________
>>> > Opensim-dev mailing list
>>> > Opensim-dev at lists.berlios.de
>>> > https://lists.berlios.de/mailman/listinfo/opensim-dev
>>> >
>>> >
>>> _______________________________________________
>>> Opensim-dev mailing list
>>> Opensim-dev at lists.berlios.de
>>> https://lists.berlios.de/mailman/listinfo/opensim-dev
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Opensim-dev mailing list
>>> Opensim-dev at lists.berlios.de
>>> https://lists.berlios.de/mailman/listinfo/opensim-dev
>
>
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>
More information about the Opensim-dev
mailing list