I agree that resources are limited, and by than I mean humans who can code. That's the largest issue we face (no matter what we're talking about). But I never said anything would happen over night. But if OS Uru ever really happens (?), then it's not a bad idea to discuss our options. And of course getting a consensus in this community can be a bit like herding cats... lol
As far as rebuilding an engine goes, again, like I said before, leveraging gives us the best benefit of our time.
Right now Plasma 2 has it's own rendering component (you know, all that stuff Brice Tebbs did). If we want to add something like HDR (which correct me if I'm wrong, but Hex Isle does have this) we'd have to integrate that into Plasma. And if we want better shader support, or shader materials we'd need to integrate that into plasma. And with any other thing like that (animation systems, compositor system..etc.) we'd have to hope to god we can make sense out of plasma, and that cyan was diligent about documenting stuff, that it's not all spaghetti, and that it doesn't give throw a ton of obstacles in our way when we try to including new things like that (or any future advances..). We'd have to integrate those things ourselves, document it ourselves, and maintain it ourselves. Indefinitely.
Create a new system with a rendering engine, like Ogre, and guess what, it comes with all that stuff, plus plugin architecture, plus plugins like paging geometry or octree scene managers, tons of third party libraries (like caelum and hydrax). Other people are writing it and using it. It's stable, and is not going to disappear. It has documentation, and it's ready to go. And it will be updated with modern features as time passes.
We will likely be spending time on plasma's architecture itself as well to make sure that physics and networking code are working correctly, maybe swap out subworlds for something better, or more sensible. Maybe mess with how Uru deals with messages, and scripting. Adding video textures, and integrate theora and ditch bink. The list of modifications to plasma continues....
Create a new system using a pre-existing framework, like IndieZen, and guess what, the component architecture would allow us to change anything from the render engine to the physics engine, to the sound engine, whenever we felt we needed to. Again, other people are writing, maintaining and using it. It's has a generic architecture, but it's designed to be capable of working for an MMO. It already integrates a bunch of 3rd party libraries (like ogre, openal/fmod..etc), so we wouldn't necessarily have to do that step either. It already has support for scene formats from ogre, and will be able to use Ogitor as an scene editor (again, leveraging another team of people's work).
Really, all we would have to do is integrate python scripting into Indiezen (basically an interface between lua and python), and make an offical format for both maps and scripting if we felt that the existing .age and .paks weren't good enough. Write some conversion programs for .ages to get the assets into a working format for the new engine. And a good converter /exporter for blender (pyprp needs to be re-written for Blender 2.5 anyways). The rest is making sure the network code works, which is the primary thing I would think we'd end up cannibalizing from plasma. Sure it would not be as simple as I just made it sound. But neither is adding all those nifty features into Plasma as it is.
And again, the key would be to design an extensible system. It would not have to be created over night. It just has to have the correct architecture that doesn't wall off our options, and makes adding things in the long term an easier process, and a process that we actually know something about.
Maybe I'm totally whacked here, but I don't think it would be so impossible to accomplish. Anyone who thinks what I'm suggesting is pure quackery (again, cause I'm not really a programmer), please feel free to correct me.
And personally I'd rather build a new house then renovate an old one that looks like it will be a b**** to fix up. Especially when the new house design is bigger and better than the original's, where as when you renovate, it will rarely be bigger or better, and any such changes would be awkward, and more difficult.
But I agree, this is something for the actual coders (who do the actual coding) to decide, if they don't already think I'm insane for suggesting this..