That's not the reason. There are 3-4 primary reasons:
- J-Selby, the author of the code doesn't feel it's ready yet
- I'd veto it because we currently don't support any basic texture functionality in the Software Rasterizer (texture interpolation for example)
- I'd veto it because we currently don't support any basic texture functionality in general (mipmapping + cubemaps)
- Texture forwarding (which this strongly depends on) still has some unsolved issues
Building software is like building other things too. Only because it's not physical doesn't mean it's simple.
It's like putting all the super-fancy furniture in your house while it's being build and doesn't even have a roof yet.
Sure, you can do that - but your furniture will soon look like shit with dirt all over it and the house will probably never be finished because construction workers have to walk around stuff which slows them down (or they might get pissed by your "I'll add furniture now lololol" attitude). What would likely happen is that they'd just throw the furniture out so they can actually work.
The same happens for Citra master / nightly: features we added to early are being kicked out / broken again + we do more important things (roof) first.
Bleeding edge is essentially cloning the work-in-progress Citra house and then a truckload of random furniture you bought of craigslist will be added: Some things will suck, some will be great - you'll have to drop some of it (because you have like 3 couches but you can only keep 1 + the chairs you like so much might not fit under the awesome table you also have = conflicts).
However, while the Citra house is still work in progress, you certainly won't buy a jaccuzi or some shit (a hose will have to do).
And texture filtering is certainly a jacuzzi.
Thanks for explanation