[sldev] What's on my plate

[Dzonatas]

Here is a quick status with everyone, so they know where I'm at lately:

...from the bottom-up:

* SCons scripts: rewrite, integration to building everything (libs, 
statics, newview, servers, etc) - reason: part of previous project but 
now needed to overcome ISO compliancy with builds that'll help solve 
issues we've seen exploited by Vista.  ISO compliancy is also needed for 
plug-in APIs else they fail like we have seen.  This is tediously taking 
up lots of my time right now to test builds.

* Sandbox builds: QA & repository. The plans with the scons scripts are 
to help with this process, so that everybody can easily work on the same 
branch or at least make cloned builds.

* Binary plug-in: Evolve LLDynamicSharedObject with current roadmaps for 
low-level APIs. This is not the same as the higher-level plug-ins, but 
"binary plug-in" is an appropriate name for the feature.

* llviewersse2: the first binary plug-in (besides llkdu). Holds various 
optimized classes. Right now just llviewerjointmesh, but I already have 
plans for a new OpenJPEG interface.

* OpenJPEG: Separate from SSE2 optimizations, I plan to experiment with 
Intel's Threading Building Blocks here to allow the tiles to decode 
under multiple threads and benchmark the results

* Sculptie importer: VWR-1110 is the start but this will move to a 
binary plug-in. There are too many features in demand, so I want to 
quickly evolve this to use scripts instead of being hardcoded as just a 
binary plug-in. That way others can extend the features easily.

* Shadows: Instead of jumping right into a real-time renderer, allow a 
feature to bake textures in-world, first. This will probably be tied 
more with the sculptie importer as scripts.

* Real-time renderer: a binary plug-in yet to be started but with TBB 
this makes a great way have scenes beyond simple lighting and shadows. I 
have a few open source models that I have analyzed and to follow foot. 
Even if the real-time renderer I make is not widely used, it will help 
evolve the plug-in layer for other renderer options or at least for 
those more dedicated to graphics programming than me.

* Client-side Scripting Engine: Python would solve the immediate need 
plus Blender is already integrated with python. I have my own VM that I 
would love to evolve further in-world and have been working on it slowly.

See my wiki page for other bits. There is a lot more I want to do, but I 
don't want to appear so random like a project whore.

Cheers!

-- 
Power to Change the Void