[X3D-Public] Fwd: Re: [X3D] X3D HTML5 meetingdiscussions:Declarative 3D interest group at W3C

[Joe D Williams]

> not about plugins

 I think I agree except in the near term I see an opportunity to offer 
some level of speedup by offering an opertional 'engine' that can 
speed things up by doing some details in fast code. In many examples 
of WebGL I have seen so far when there is a lot oc computing then the 
thing is typically slow like before I got my first real video card 
with VRML2. I just remember how that made it real for me and got me 
wanting to learn about it. Similarly, I saw some great speedups in X3D 
when we could get rid of scripts with stock utilities to do some 
simple processing. I mean I am sure we have learned by actual 
experience how nice it is to have some interpolators and sequencers 
and sensors and detail utilities running in the 'engine' core than 
from any collection of script libraries that somehow work to time 
things in an environment where Time is not a first class object.
I may be way behind in what is happening but I have seen the speedups 
in X3DOM when we hook in pieces of a majik little engine and some 
results when we tried to eliminate the special 'engine' and do 
something interesting without some neat 'engine' hooks.
To me, this means to do good either the browser must buildin some 
power or add a simple 3D engine to do some of the neat stuff. We have 
already heard about how we could get isOver and isActive from a set of 
onmouse events and I think inferred that some special code somwhers 
would do some neat raycasting and produce events.
So, I think both solutions 'need' a concept that is a plugin but is 
not a plugin. I am also very impressed with integration of MathML and 
SVG, and even Canvas2D. Another consideration is that work is still 
proceeding on developing the html <object> and <embed> protocols and 
rules.

If this works as hoped for, then it will be much easier to get a 
'plugin' installed and working. We saw from Ajax3D that XHR is a 
winner and so naturally would play some part in future X3D networking 
tactics. Now it seems the html browsers are all up in the air about 
security of <object> and @sandbox and others so we need some special 
tricks to get the X3D object talking to the DOM. Like Tony's old 
Ajax3D examples stopped working and I naven't taken the time to get 
them to run again. A great use of how to use host DOM and html browser 
networking features. Of course the XHR worked better when the X3D 
could deal with the network directly instead of through the host DOM.

Anyway, plugin or not, it seems clear that some 'engine' code will be 
required. Either the host will provide suitable capabilities or a 
little addin code is provided. I think we can tell a lot about how 
appropriate this 3D offering will be by examining details about how 
this thing will work when it is a live context in a host DOM and being 
kicked from the inside and the outside while responding to style and 
content updates.

Of course we need to know something about the data syntax, structures, 
and features to do this, but since we seem to have a decision that 
whatever language will be hierarchal allowing nested transforms and 
that the encoding must produce a valid DOM, then we can go ahead an 
make some plans about how this 3D DOM which may be a distinct context 
embedded in an html canvas that listens and talks to our WWW via the 
intermediary of our very much appreciated personal favorite host html 
text browser.

Finally, this is not just about html, I hope, but can emcompass xml. I 
mean we need XML tools to work with this stuff so we also need a 
validating schema and XML DOM.

Thanks to All and Best Regards,
Joe

----- Original Message ----- 
From: "Chris Marrin" <chris at marrin.com>
To: "Joe D Williams" <joedwil at earthlink.net>
Cc: <x3d-public at web3d.org>
Sent: Sunday, January 02, 2011 5:35 PM
Subject: Re: [X3D-Public] Fwd: Re: [X3D] X3D HTML5 
meetingdiscussions:Declarative 3D interest group at W3C



On Jan 2, 2011, at 4:00 PM, "Joe D Williams" <joedwil at earthlink.net> 
wrote:

> Hi All, when thinking of somehting new, please think about mime 
> types for the main text file and maybe even files that are optimized 
> in relation to the data or intentions you are moving around.
>
>>
>> Thanks Again and Best Regards,
>> Joe
>> http://www.hypermultimedia.com/Logo/Web3dLogo-X3d-animated-logo2.x3d
>>
>>
>
> believe it or not:
> I can click that link in my email Outlook Express and get IE8 with 
> BSContact installed and it actually works. Same for .x3dv and for 
> .wrl.
> That also happens in most all web browsers clicking an html link 
> when the plugin is installed and the mime is recognized.
> This is happening due to solid registered (at least in the host 
> browser) mime types
> More and more of the web is depending upon mimes, believe it or not.

If the format we're discussing is declarative 3D markup in a web 
browser, we already have he mime- type: text/html. I don't say that to 
be trite. It's an important requirement. Modern HTML 5 parsers deal 
with embedded markup very well. It let's you embed SVG, MathML, etc. 
inline. There are beautiful examples of embedded MathML that work with 
most of the bleeding edge browsers. IMHO thinking of this as a plugin 
is a non-starter.

>