[x3d-public] Fw: Question about MU and Network Sensor

Sun Apr 21 12:30:06 PDT 2019

I was hoping you could provide code snippets showing an example of this implementation, as going through and dissecting your entire project to find it myself is a little out of reach time wise.

Buffering, as in consolidating events from the various NSs into a minimum number of connections, as opposed to initiating a connection every time an event calls for it. 

I talked about the two sides of a coin, but  you may or may not have visibility to another aspect of what a NetworkSensorNode is; namely the browser side or, the client side of the client-server relationship. The NetworkSensorNode is the interface between the scene and the network, and if we only looked at scene authoring, then we'd be completely missing out on the network side, which is really what's important here.

Specific implementations may have different goals and objectives, but I am not too concerned with that. I'd like us to focus on the network connection interface directly, in a way that it can be generalized and applicable to a wide range of applications. 

GL   

From: x3d-public [mailto:x3d-public-bounces at web3d.org] On Behalf Of Christoph Valentin
Sent: Sunday, April 21, 2019 2:46 PM
To: x3d-public at web3d.org
Subject: Re: [x3d-public] Fw: Question about MU and Network Sensor

>>>>> However, once you do buffer events in order to manage connections, then the perceived benefit of placing NetworkSensors inside models may become moot, though perhaps under specific scenarios an author may elect to use that approach. Have you tried such an implementation?

I did not understand the issue with the buffering ?????? Is it about the HNSI?
-- 
Diese Nachricht wurde von meinem Android Mobiltelefon mit GMX Mail gesendet.

Am 21.04.19, 19:59, Christoph Valentin <christoph.valentin at gmx.at> schrieb:

>>> Have you tried such an implementation?

Yes

http://simulrr.sourceforge.net

>>> I do not really see how it simplifies authoring, as models portability would also be impeded.

If it's agreed as an ISO standard, the portability would even be increased, avoiding museum like 3D models

>>>Perhaps a specific example would help.

http://simulrr.sourceforge.net

Gesendet: Sonntag, 21. April 2019 um 19:13 Uhr
Von: "GL" <info at 3dnetproductions.com>
An: "'x3d-publicweb3d.org'" <x3d-public at web3d.org>
Betreff: Re: [x3d-public] Fw: Question about MU and Network Sensor

Absolutely. In its most basic form: X3DNetworkSensorNode = Connection + NetworkSensor(s)

You are essentially asking if it is ok to separate the two sides of a coin,
and "bury/hide" one side into models. I get that. The problem becomes one of connection instantiation. Unless you have in place a mechanism for buffering events, the NetworkSensors will keep on instantiating connections. However, once you do buffer events in order to manage connections, then the perceived benefit of placing NetworkSensors inside models may become moot, though perhaps under specific scenarios an author may elect to use that approach. Have you tried such an implementation? I do not really see how it simplifies authoring, as models portability would also be impeded. Perhaps a specific example would help.

GL

From: x3d-public [mailto:x3d-public-bounces at web3d.org] On Behalf Of Christoph Valentin
Sent: Sunday, April 21, 2019 6:08 AM
To: x3d-publicweb3d.org
Subject: [x3d-public] Fw: Question about MU and Network Sensor

I have not yet answered the part of the example about entering the car and driving away........

Let's first discuss the basics, OK? :-)

Gesendet: Sonntag, 21. April 2019 um 11:47 Uhr
Von: "Christoph Valentin" <christoph.valentin at gmx.at>
An: "x3d-publicweb3d.org" <x3d-public at web3d.org>
Betreff: Re: [x3d-public] Question about MU and Network Sensor
Gina-Lauren, All

To stick to your example: No, I am not suggesting that. You are bringing two things together that are not related to each other.

1. Having Network Sensors "buried" within models
2. Each Network Sensor using their own network connection

These two things are not related with each other. If you had had a look to my drawing, you would have seen the network connection is maintained by the "Network Sensor Infrastructure" and all "Network Sensor"s use the same connection, using the "Hidden Network Sensor Interface" (HNSI).

Getting back to the example:

The hat of the avatar and the bag he is carrying are two examples of models that don't need any synchronization. Given the hat and the bag are rigid bodies, they are just rendered relative to the hand of the avatar and relative to the head of the avatar.

The avatar has to care for the synchronization of the position of his hand and of his head (and therefore of the bag and of the hat).

The car and its doors, the steering and the drive, are a perfect example for what I mean, on the other hand. In my humble opinion, the model of the car should already contain all environment sensors (e.g. a touch sensor for each door) and all network sensors for the synchronization of the car (of its doors, steering and drive).

The surrounding scene should as little as possible be affected by the following facts
• it is a car, that is attached to the scene
• the car is MU capable
• the specific type of the car (racing car, offroad car, ......)

So let me rephrase my question to x3d-public:
a) Is it OK to require from each Network Sensor implementation to use a HNSI for the connection of all Network Sensors to the Network Sensor Infrastructure, in order to ENABLE burying of network sensors within models?
b) Is it OK, to ALLOW model authors to bury their Network Sensors within their models without bothering scene authors too much?

One detail: we are using the term "network connection". This term is poorly specified, but I think we can agree to use it, as long as we define we are meaning one of the following things:
• TCP connection
• SCTP association
• SIP dialog
• .......and so on

All the best
Christoph

Gesendet: Sonntag, 21. April 2019 um 07:25 Uhr
Von: "GL" <info at 3dnetproductions.com>
An: "'x3d-publicweb3d.org'" <x3d-public at web3d.org>
Betreff: Re: [x3d-public] Question about MU and Network Sensor
Christoph,

Let's say for example that you have an avatar wearing a hat and carrying a bag. That avatar opens a car door, gets in and drives away.

Are you suggesting that the avatar, the hat, the bag, the car door and the car itself should all have a separate NS within and make each their own connection to the server? It is feasible, but unless I am missing something from this reasoning, my take on it would be that the server would get overloaded rather quickly, resulting from all of the extra connections. So in other words, instead of being able to handle say 50 avatars per machine, now we'd have only 10 (in this example), since each model would have to synchronize via their own connection. Also, network latency could cause the hat to fall behind or the car door to get ahead of the car and everything.

Like I've mentioned before, I could see something this working somewhat better with Java3D (though I haven't touched it in years), but with X3D it brings up questions to my mind. So to answer your question, IMO it is better to have fewer connections and pack as much info into them as possible. But assuming that you'd want to do it anyway, the models would still work and interact with the scene as specified (however we're getting into other considerations here that exceed the context of this post).

GL

From: x3d-public [mailto:x3d-public-bounces at web3d.org] On Behalf Of Christoph Valentin
Sent: Saturday, April 20, 2019 12:26 PM
To: x3d-publicweb3d.org
Subject: [x3d-public] Question about MU and Network Sensor

Dear all,

Am trying to figure out one detail about the highly discussed "Network Sensor".

Have now explained the "operational paradigm" of the network sensor, as I think to understand, at our SP-ARK WIKI:

https://github.com/christoph-v/spark/wiki/Operational-Paradigm

If you follow that link, you will find I am trying to use my own words in order to describe the well-known. This is not only due to the fact that I have got a fatal tendence to invent the wheel another time and another time, it it also due to Copyright considerations.

If you follow the link, then you will find my impression the interface of the Network Sensor is twofold.

First there is the what-I-call CNSI (Common Network Sensor Interface).

Second there is the what-I-call SNSI (Specific Network Sensor Interface).

Now getting to the point: is it acceptable to require the SNSI MUST remain WITHIN the model? Is it allowed to have it affecting the "master scene"?

I think this is an important question, otherwise I wouldn't bother you.

All the best

Christoph

P.S.: Currently the WIKI has got only two pages, the home page is https://github.com/christoph-v/spark/wiki

_______________________________________________
x3d-public mailing list
x3d-public at web3d.org
http://web3d.org/mailman/listinfo/x3d-public_web3d.org

_______________________________________________
x3d-public mailing list
x3d-public at web3d.org
http://web3d.org/mailman/listinfo/x3d-public_web3d.org

_______________________________________________ x3d-public mailing list x3d-public at web3d.org http://web3d.org/mailman/listinfo/x3d-public_web3d.org 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20190421/a97a4aef/attachment.html>