I would say what I should do from the start of this compiler run, instead of running it on normal classes with various different situations which was a huge mistake and error to do... I will instead build a bunch of basic sample classes with a single set of scenarios that I can implement and test for accordingly. Hopefully with these scenarios it will be easier and a more stable implementation with less guess work.
I actually got rid of
Poking.poke() which makes things nicer. I do have
another good idea and that is for TAC to have a potential alternative main
for some tests, so if I want to test the interpreter I can by just having a
slightly different main starting point.
Okay so I added support for proxy main in tests. I think what I should do next is implement a bunch of tests around the interpreter and also adding basic implementation for it just solely for tests. This way I could potentially actually run the target tests via the interpreted environment. If successful I would be one step closer to having SummerCoat be supported but also have it be a bit faster for stuff that is native level. So in the future when graphics functions are actual native code, they would not be as slow as a pure interpreter.