[Opensim-dev] IRegisterInterface for extending scene entities

Fri Jan 18 23:46:57 UTC 2013

Alright, I'm going to suggest

1.  That all attributes must be within their own OSDMap, which must be a minimum of 4 chars long (maybe this could be 
relaxed for 'core' stuff).

2.  Within this map, attribute names and further OSDMap or other structures are entirely up to the caller.

3.  At the initial development stage, everything may change such that existing data may no longer be retrieved (e.g. the 
way this is stored or the entire approach may change).  One must not rely on data being present.

4.  If at all possible, this must be backward and forward compatible with other versions of OpenSimulator no matter 
whether it remains or is altered radically.  This should be fine since deserialization will ignore any top-level xml 
nodes that it doesn't recognize.  But we must preserve a Postel's law - this was a painful lesson to learn.

On 18/01/13 22:09, Adams, Robert wrote:
> I have a need to add more attributes to an SOP (use specified center-of-gravity, user specified mass, rolling friction, ...) and the implementation of Justin's dynamic attributes is just what I need. It does not fulfill my other wishes of adding functionality to SOPs, but it does allow the dynamic addition of new and serialized objects attributes.
>
> Can we Just Do It(r)(tm)(sm)? Some comments in the code suggesting a top level naming convention ("prepend attribute names with your module name" or some such) and an admonition to not store large things might be sufficient to control the use of a simple OSDMap.
>
> What do you say? I am +1. How about everyone else? It is a start.
>
> -- ra
>
> -----Original Message-----
> From: opensim-dev-bounces at lists.berlios.de [mailto:opensim-dev-bounces at lists.berlios.de] On Behalf Of Justin Clark-Casey
> Sent: Friday, January 04, 2013 5:18 PM
> To: opensim-dev at lists.berlios.de
> Subject: Re: [Opensim-dev] IRegisterInterface for extending scene entities
>
> No problem in making this e-mails, Robert - I think this is exactly the kind of discussion that we need.  These are the kind of decisions that we'll be stuck with for a long time so I think we need to think them through, both through discussion and review or proof-of-concept code.
>
> I hadn't really properly considered that the JSONStore is close to this.  I suppose one other significant difference is that the current dynamic-attributes effectively stores data per part rather than through a single datastore for the whole region.
>
> But even if it does use the blackboard pattern, there still has to be an underlying persistence store, which in this case would be a serialized JSON object that is functionally identical with the OSD Map (I've run out of time today to see if the JSON store enforces type safety in some way).  Really, the code could be very similar, though in terms of formats I would prefer if to avoid further schizophrenia over data storage in OpenSimulator.  If we stored as JSON, for instance, we end up transporting a JSON object in XML... :p
>
> On 04/01/13 21:22, Adams, Robert wrote:
>> I think adding a module allowing scripts to attach dynamic variables to SOPs is the right level of functionality. Just adding a key/value store is way too low a level.
>>
>> For instance, the JSONStore module (http://opensimulator.org/wiki/JsonStore_Module) creates a dynamic variable storage to share between scripts. Since it's made for sharing, a script can request events when values are updated. The module adds, as one piece, functionality for extension and communication between scripts.
>>
>> Does this attribute store enable such? Would a script want to know when one of the key/value pairs changed? Is polling suitable? Does the usage case just need static variable addition?
>>
>> As to the practical, implementation angle, yes, you are correct that adding an OSD store on a scene object is simpler. This is an open source project, after all, so this is just my opinion.
>>
>> -- ra
>>
>> -----Original Message-----
>> From: opensim-dev-bounces at lists.berlios.de
>> [mailto:opensim-dev-bounces at lists.berlios.de] On Behalf Of Oren
>> Hurvitz
>> Sent: Friday, January 04, 2013 11:22 AM
>> To: opensim-dev at lists.berlios.de
>> Subject: Re: [Opensim-dev] IRegisterInterface for extending scene
>> entities
>>
>> The typing problem is already solved since the dynamic attributes are implemented using OSD.
>>
>> Your proposal is much more complicated than the currently proposed
>> attributes, and I don't want to suppress a good solution in favor of a
>> perfect solution that might never be implemented (I assume you're too
>> busy with BulletSim to do this...) Furthermore, it would prevent a
>> very exciting
>> use-case: the ability to use dynamic attributes from a script, using OSSL.
>> (Well, unless you add a module for that specific purpose, but in that
>> case you just added a layer of indirection around the key-value
>> store.)
>>
>> Oren
>>
>>
>>
>> --
>> View this message in context:
>> http://opensim-dev.2196679.n2.nabble.com/IRegisterInterface-for-extend
>> ing-scene-entities-tp7578406p7578425.html
>> Sent from the opensim-dev mailing list archive at Nabble.com.
>> _______________________________________________
>> 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
>>
>
>
> --
> Justin Clark-Casey (justincc)
> OSVW Consulting
> http://justincc.org
> http://twitter.com/justincc
> _______________________________________________
> 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
>

-- 
Justin Clark-Casey (justincc)
OSVW Consulting
http://justincc.org
http://twitter.com/justincc