[openvrml-develop] Priorities for 0.17

[Braden M. <br...@en...>]

Now that 0.16.0 is out the door, I'd like to firm up priorities for
0.17.

As far as I'm concerned, the "must have" feature for 0.17 is the
partitioning of the node implementations and scripting backends into
separate modules. This will require relatively few code changes since
the last few major releases of OpenVRML have laid the groundwork for
this. For 0.17, it's finally time to make the nodes-as-components a
reality. (See tasks 68361 and 34921.)

Other relatively high-priority items that I hope to make the cut:


  * Bring the Script node Java binding (back) to a basically workable
    state. (Bug 1099810 and more) 
  * Encapsulate display list/texture object caching in the renderer.
    (Task 126915) 
  * Text rendering for Windows. (This might be workable for a 0.16.x
    release.) (Task 86860)

But in the interest of having releases more frequently than annually,
I'm not inclined to hold 0.17.0 for any of the above three items.

There are also a couple of "playground" items that don't dramatically
impact functionality, but are of interest to me because I enjoy working
on them:

  * Replace the Antlr parser with a Spirit one. 
  * Figure out how to employ template metaprogramming to dramatically
    reduce the volume of code needed to create a concrete node_type
    instance. 

A more vague item I'd like to pursue soon is "make navigation not suck".
Apart from the existing prerequisite tasks to making that happen (tasks
18156, 18157, and 18154), my plan for this is to provide a framework
such that navigation modes (i.e., NavigationInfo "types") can be created
using VRML and scripting. I would like to provide tools to make OpenVRML
a useful platform for experimenting with new interface styles; in
particular I don't want to restrict users to the navigation modes
specified for VRML97 and X3D (which are adequate but rarely ideal for a
particular world).

Making this happen will involve:

  * introducing a notion of a separate scene graph branch where the
    navigation type implementation nodes live; 
  * implementing nodes that provide more control over the user view and
    interaction than the ones in the VRML/X3D node set. 

While I don't think any of this can be fully realized in the time frame
in which I'd like to make 0.17.0 happen, I do hope to be able to lay
enough groundwork that it's realistic for 0.18.0.

Finally, there are a couple of high-value items that do not involve C++
skills, for anyone who might want to contribute but not know C++:

  * Write a comprehensive test suite for the Script node JavaScript
    binding. 
  * Run 0.16.0 through the conformance tests to see where we stand.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>