[x3d-public] Open Asset Import Library download | SourceForge.net

[Joe D Williams]

> Q1. how is x3d -or xml, or any branchy tree data- stored in a 
> database?

In the case of X3D, it is the database. Yes, the X3D scenegraph is the 
database. And, Yes, it is alive!

Why consider any other database style than X3D XML encoding? It can 
produce a validatable n-D authortime and runtime database that can 
leave realtime tracks in any dimension and be used to interactivly 
vizualize the present(s) and simulate futures. If some part of your 
feature names and data needed to store or produce your effects is 
missing, then there is plently of room to create your own meta 
ontology and data and events -- even interact with them. Then, you can 
script it while it is running to keep it up to date or whatever, and 
produce a serialization at any moment.

So, to interact with the data at authortime all you need is XML/DOM 
tools that like X3D XML Schema.
To fully vizualize and animate it, you need a time- and event-driven 
player that likes X3D events.

All Best,
Joe



----- Original Message ----- 
From: "doug sanden" <highaspirations at hotmail.com>
To: <x3d-public at web3d.org>
Sent: Wednesday, May 27, 2015 9:29 AM
Subject: Re: [x3d-public] Open Asset Import Library download | 
SourceForge.net


OK thanks again John, I got all that, I'll save it for when I do a 
project, thanks very much.
-Doug
> John Carlson, on web3d public
>> John,
>> I'm not a database guru, but I've always wondered:
>> Q1. how is x3d -or xml, or any branchy tree data- stored in a 
>> database?
>> -- a) an entire scene as a blob?
> Generally in Oracle it's stored as an xml type so you can do XPath 
> in sql. I don't know implementation details. I suspect serialized 
> DOM to blob however.
>> -- b) each node type in a separate table, with columns representing 
>> fields (about 256 node types->256+ tables?)
> Try to come up with a SQL query to output XML and you will see why 
> other means are chosen.
>> -- c) few more general tables like nodes, scalar fields, MF fields, 
>> parentNode+field/child, parentMFField+index/child
> We did something like this with HDF5. Essentially you create tables 
> around the document and then use ETL to populate the tables.
>> - is there a way to generate a table layout automatically using 
>> XSLT for x3d?
> I'm not sure why you would want to when you can use NoSQL. No tables 
> necessary.
>> Q2. how is the parser/loader generator related to ETL? (I see ETL 
>> often, I think it means extract, transform, load)
> ETL breaks up documents into relational form.
> Generally branchy trees are stored in the database as JSON these 
> days I'd say. But I only have experience with CouchDB and MongoDB.
> And it should be fairly simple to come up with an array table, an 
> object table and an attribute table. Querying is going to be a bear 
> unless views are used though.
> Realize that arrays are objects too (with integer attribute names). 
> That reduces it to object and attribute tables.
> And probably anything can be stored with (objectId, attributeName, 
> attributeValue) tuples. I might add revision to that.
> Oracle has a way of querying hierarchical tables...CONNECT 
> something.
> It looks like the SQL equivalent to Oracle's CONNECT BY is WITH.
> the other thing you do with HDF5 is extract images to a file system, 
> then store the filename in the database.
> Last version of HDF5 I used only did single threaded access to the 
> file in Java. That was one main reason why we did ETL.
> https://www.hdfgroup.org/HDF5/
>



________________________________
> Date: Wed, 27 May 2015 11:23:06 -0500
> Subject: Re: [x3d-public] Open Asset Import Library download |
> SourceForge.net
> From: yottzumm at gmail.com
> To: highaspirations at hotmail.com
>
>
> And probably anything can be stored with (objectId, attributeName,
> attributeValue) tuples.  I might add revision to that.
>
> On May 27, 2015 11:17 AM, "John Carlson"
> <yottzumm at gmail.com<mailto:yottzumm at gmail.com>> wrote:
>
> It looks like the SQL equivalent to Oracle's CONNECT BY is WITH.
>
> On May 27, 2015 11:10 AM, "John Carlson"
> <yottzumm at gmail.com<mailto:yottzumm at gmail.com>> wrote:
>
> Realize that arrays are objects too (with integer attribute names).
> That reduces it to object and attribute tables.  Oracle has a way of
> querying hierarchical tables...CONNECT something.
>
> On May 27, 2015 11:02 AM, "John Carlson"
> <yottzumm at gmail.com<mailto:yottzumm at gmail.com>> wrote:
>
> On May 27, 2015 10:36 AM, "John Carlson"
> <yottzumm at gmail.com<mailto:yottzumm at gmail.com>> wrote:
>
> > Generally branchy trees are stored in the database as JSON these 
> > days
> I'd say.  But I only have experience with CouchDB and MongoDB.
>
> And it should be fairly simple to come up with an array table, an
> object table and an attribute table.  Querying is going to be a bear
> unless views are used though.

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