Re: Is Sphere less of an engine and more of a standard now?
Reply #1 –
Probably not, no.
tl;dr: The answer is to use minisphere, unless you know exactly why you shouldn't use it. All the other implementations either died out or are now not really Sphere compatible. minisphere is probably the only engine anyone who just "wants to use Sphere" should use, or maybe 1.5/1.6 in certain circumstances (since they still work).
Sphere 1.5/1.6 is still preferable for very, very obscure platforms.
TurboSphere is kind of a dead end for Sphere-related games. It implements the entire map engine, but other than that its API is now not very similar to Sphere's, since I let it drift so far with the promise that I'd write a compatibility shim later. "Later" probably means never at this point. This is compounded with there only being a Mac OS X and a Solaris port anymore, so 90% of everyone can't use it, and there is no editor for either of those platforms.
However, given that minisphere implements the Galileo API from TurboSphere, and contains almost all of the 1.5 API, it has most of the benefits of TurboSphere (unless you are doing some really crazy stuff that needs ultra-speed, or you need ArrayBuffers (question mark?)), and is fully compatible with Sphere 1.5.
Using Sphere 1.6 in combination with FJ-GL will give you close to the same performance as minisphere, but there are a few cases it still pulls away, and the difference will probably become greater and greater as Allegro and Duktape keep getting upgraded and SpiderMonkey 1.7 remains the same.
I'm not sure where Sphere-sfml ended up, but I believe it's at "mostly works with the 1.6 API". However, its JS engine is Windows only, and probably will stay that way for a long time. I wouldn't hold my breath.
Sphere 2.0 ended up the same as TurboSphere in a lot of ways, and I totally see how and why Kyuu took it the direction he did now. It's no longer compatible with Sphere (even less than TurboSphere, where you can still kind of fake it since it still uses JS).
This is, as far as I am concerned, the endgame of Sphere 1.6 and the original engine. No one is really interested in upgrading its JS engine, and my experience with FJ-GL proved to me that the actual engine design has reached its limits for performance. Additionally, the engine is fairly complex, and no one is really left who understands it enough and/or cares enough to try to actually modernize it. I tried once, but I believe it's not worth the time. It would be easier and better to just start from scratch, or use an existing engine (minisphere or TurboSphere) as the basis for any improvements.