Well, i did re-think this stuff, more or less. What do you think about the scheme below, guys?In fact i'm now trying to re-think the whole adaptation and money spending algorithm. The problem with NullBot3 (which was not showing up in a simplier NullBot2) is when he knows exactly what he needs to spend money on, but is unable to, for various reasons, and then he needs to decide whether to spend money on something else or save money until the object becomes available.
Do you see any obvious bugs?
tl;dr: the main difference between this scheme and NullBot (all incarnations) is:
- No longer do "Make everything occupied, it will queue up money automatically." If we don't have money, do not spend it.
- No longer do "First find idle lab, then fill it with something". Research has higher priority than building/producing. But we keep only a fixed amount of labs active, depending on our oil income; idle labs will remain idle if we do not have enough oil.
- No longer do "First find idle factory, then try to make sure it produces something". First of all decide, out of all buildable/producible objects, what do we need the most. Only then build/produce anything at all. Consider tower construction and droid production together rather than separately.