Re: [ogpx] Draft work on Foundation and Type System

[David W Levine <[e-mail]>]

On the type system, I'm glad to see the LLSD serializing to other 
transports getting some love. My sense is that there are, effectively four 
types of things which cover the space. The bulk, as you note is
RESTful http semantic operations.  There are two other common types of 
communication which flow today, and which need to be captured as we move 
forward. There are updates which, if we *could* 
expose a REST resource on the client, we would post directly to the 
client. But these do have the flavor of being fundamentally unsolicited. 
While one can model these as a series of "posts" against the
general client, its not clear to me that this fully captures what's 
happening, nor does it help one associate the updates with the source in a 
clean fashion. There is the question of whether *some*
of the current UDP traffic "Take two steps forward" "Rotate this prim 5 
degrees" again models exactly as posts on a REST resources or something 
different.  There is also a small set of operations which
probably fall close to the "certified http" model Which Linden was working 
on a while back, where there is a very strong need to ensure close to ACID 
behavior.

 I'd love to see as much of this space  cleanly modeled as LLIDL services 
as possible. I don't imagine it all being solved in one draft, one of the 
strengths of the IETF model is being able to issue
successive refinements to a draft as the community's  understanding grows. 
 

I firmly prefer the core LLSD/LLIDL stuff in one draft.  I think it *may* 
make some sense to have a "this is baked" draft and a "This is being 
worked out" informational draft which will feed into the baked
draft in the future. 

- David
~ Zha




From:
Mark Lentczner <[e-mail]>
To:
[e-mail]
Date:
03/01/2010 01:42 PM
Subject:
[ogpx] Draft work on Foundation and Type System
Sent by:
[e-mail]



I'll be working on another revision to the Foundation and Type System 
drafts this week.

I see the following issues that I'll be taking on:

1) Type System: Missing comment syntax -- I'll add it

2) Type System: Binding to HTTP -- My approach will be to reword this so 
that LLSD could be serialized to other systems, and LLIDL described 
services could easily be bound to other transports. However, the abstract 
model that they from is very REST like, and REST as embodied by HTTP. So, 
for example, the existence of a small number of transport "verbs" is 
assumed, whereas the availability of headers is not. I do think, like 
Josh, that since the vast bulk of use is indeed over HTTP, that splitting 
into multiple documents doesn't serve at this point. Remember: Future 
drafts are perfectly capable of referencing only parts of other drafts.

3) Foundations: Continued word smithing.

- Mark Lentczner

Mark Lentczner
Sr. Systems Architect
Technology Integration
Linden Lab

[e-mail]

Zero Linden
[e-mail]




_______________________________________________
ogpx mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/ogpx