[Opensim-dev] OGPX and IETF-ing things

Thu Jun 4 02:39:50 UTC 2009

Thanks for the clarification, Meadhbh. And sorry, I also couldn't parse 
the gender of your name :-)

In general, I have no problem with companies wanting to make standards 
out of the things that they develop. However, we've only started to 
scratch the surface on protocols for interoperability in VWs. There 
isn't just OGP and the Hypergrid, there are many possibilities, 
especially when it comes to issues of authority (I could spend another 6 
  months designing another 2!). Wanting to make standards about that 
this early on is utterly premature.

As you already found out, there's a lot in common in the protocols that 
have the SL "genes". It would be your loss, I think, that you miss 
those, because if you go ahead with standardizing OGP without engaging 
with the several of us exploring other protocols, and then you find that 
you missed something interesting, you end up stuck with a standard that 
no one but you will use.

This is not to say that there should be only one interop protocol. My 
point is exactly the opposite, especially at this point: we need to 
explore the design space much more, see what works for what purposes, 
etc., and then, maybe, converge into the minimum common denominator. 
This can only be done through actual technical and commercial 
experimentation, a lot of it. OpenSim is absolutely great for that.

I wonder what's the appropriate medium/process for exchanging these 
technical ideas. I don't think a mailing list called "OGPX" with the 
agenda set to "focus on OGP" and with a very clear Linden Lab mandate to 
standardize the parts of the LL protocol that you already have in place 
is appropriate. Certainly the Hypergrid and all other protocol ideas 
currently being implemented (e.g. Cable Beach, MXP) don't fit in there.

I want to think that opensim-dev (the mailing list and the IRC) are 
sufficiently chaotic in a very interesting way that no one, including 
you and other LL employees, should feel uncomfortable raising any topic 
at all, and that everyone feels that most discussions are productive, 
because they usually end up in a piece of code being written by someone.

So please, do come out of your hiding place high in the clouds and ask 
questions about how the Hypergrid2 uses inventory CAPs, about Cable 
Beach's servers, about MXP, and everything in between; answer the 
unanswered questions about OGP that occasionally are raised here. [for 
example, my description that in OGP the regions still do agent transfers 
has never been confirmed explicitly, and I never read it anywhere; it's 
just my understanding of OGP. So how does that work?]

In spite of the fact that this group of people aren't the fan club that 
you usually interact with, I'm pretty sure you will find a lot of 
positive spin for LL's technical ideas here, the good ones at least.

Infinity Linden wrote:
> a couple points...
> 
> a. it's ms. hamrick, not mr. hamrick. but don't worry... i get that a lot.
> 
> b. the purpose of the OGPX mailing list is to discuss requirements for
> standardization as they exist today. with the expectation that we'll
> eventually form a working group "soonish."
> 
> c. our plan is to take the bits of a protocol we have general
> agreement on and standardize them now. once they're standardized, we
> recharter and work on the next couple of bits, and then keep repeating
> this pattern til we have something resembling a complete standard.
> 
> d. i think i'm starting to understand where Christina's coming from...
> it seems to me her approach is "code now, standardize later," which is
> a PERFECTLY valid thing to do. and i also agree that ultimately, HG
> and OGP will remain distinct protocols. but it sounds like there's
> still some things we're both using. capabilities is the perfect
> example. we're both using them, and we just had a request to add some
> form of introspection. we're probably going to add some verbiage in
> the spec that OGP servers should respond to an OPTIONS request with
> the LLIDL of the resource represented by the capability. it would be
> nice that before we put that in ink, we get someone with some
> understanding of how HG is using capabilities gives it a once over and
> lets us know whether it's going to break anything on the HG side. If
> we find out about it early, we can change it. Finding out about it
> after systems have been deployed makes it a lot harder to change.
> 
> e. yup. we like OAuth and ProtectServ (well... some of us like
> ProtectServ.) OpenID is a bit of a non-starter because the OpenID
> community has declared "authenticating a desktop application" an
> un-interesting use case. That being said... what John's doing where
> OpenID is used in a web browser, and that browser then forwards to a
> secondlife: URL or to a HTTP URL that downloads a blob with a user
> token. That's actually pretty cool. However... it is sort of outside
> the realm of OGP. Though if peeps want to work on an informational RFC
> re: best practices for integrating OpenID with OGP, that's cool too.
> If you want to do it on the OGPX list, that's fine. if you want to do
> it on sldev, that's fine.
> 
> so... to recap... OAuth definitely has a place in OGP. OpenID? sure.
> there's lots of stuff we could do there, but probably not IN OGP
> itself. OpenSocial... sure. if there's interest.
> 
> -cheers
> -meadhbh
> 
> On Wed, Jun 3, 2009 at 2:55 PM, Christian Scholz <cs at comlounge.net> wrote:
>> Hi!
>>
>> Toni Alatalo schrieb:
>>> On Jun 3, 2009, at 10:37 PM, Cristina Videira Lopes wrote:
>>>
>>>> Personally, I think this is all premature. IETF-ing the Hypergrid is
>>>> premature for different reasons than IETF-ing OGP is premature. The
>>> I think that depends on what is meant by IETF-ing. For proposing as a
>>> standard, you are most probably correct. But starting to discuss
>>> something within the IETF doesn't mean there's a ready proposal for a
>>> standard - perhaps just an idea of the need, a set of requirements and
>>> some early implementation. For example a working group can then discuss
>>> based on those.
>> If Cristina says it's too premature for a standards track then I guess
>> that's right but I also think that putting things up there maybe on a
>> separate mailing list and just discussing ideas/proposals for the
>> protocol there could be useful. There are a lot of people very informed
>> about protocol building and security issues which might be very helpful.
>> It also means that maybe other VW vendors (who produce right now even
>> more proprietary protocols) might be included at least a little bit and
>> it might lead to a better chance at some interoperability in the end
>> (where I agree there is space for different implementations but I would
>> see this more in specific areas, for some areas I still think it makes
>> sense to reuse existing standards such as OpenID, OAuth, OpenSocial and
>> the like).
>>
>> -- Christian
>>
>>
>> --
>> COM.lounge GmbH
>> http://comlounge.net
>> Hanbrucher Strasse 33, 52064 Aachen
>> Amtsgericht Aachen HRB 15170
>> Geschäftsführer: Dr. Ben Scheffler, Christian Scholz
>>
>> email: info at comlounge.net
>> fon: +49-241-4007300
>> fax: +49-241-97900850
>>
>> personal email: cs at comlounge.net
>> personal blog: http://mrtopf.de/blog
>> personal podcasts: http://openweb-podcast.de, http://datawithoutborders.net
>>
>> _______________________________________________
>> 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
> 