Been writing a NES emulator for a while. Works well enough but there is a new automated test available. My emulator does ok, but not good enough. Going to try and improve things.
The processor definition is in Scala using Chisel. This produces an HDL (what's needed for actual hardware) using a typed embedded domain specific language. This also compiles to a software model for validation. But that's not the same as "software only"
Beyond All Reason is really interesting. There is a long history there! The engine was originally made to play Total Annihilation mods (or the full game? Not sure). That evolved over something like 15 years!
The end result though is not just a good open source game but a damn good RTS with an awesome scene. I <3 BAR.
Anyways, aside from fawning over BAR what's my point?
Plan for the long game: these projects can take a long time to gain traction.
That does constrain the optimizations to a profile determined prior to the end user. Which is hopefully the same but might not be. Shared code between applications also is a challenge.
Specifically impacting poor people. Which, ya, is fucking bullshit. Feed the kids in school please. Breakfast, lunch, dinner if you must. Feed the ducking kids.
Mac UI development was the nicest last i tried. Even that has a slow dev loop compared to Web.
I did ok with GTK plus a nicely fixtured launcher. Cut the cycle time down to dozens of seconds. But still... Meh.