raymondben
commented
2 years ago Have applied a y-adjustment to both the attack code and attack player guess functions, but probably needs a better solution. Perhaps inference of attack code and player simultaneously, based on actual (continuous) coordinates for both (rather than zone for player as it is now). Perhaps also looking at attack coordinate relative to setter coordinate (when sets are being scouted), which might help disambiguate e.g. X1 from X2 from X7.
At the moment we take the clicked coordinate as the court location of the attack - so, a click just behind the 3m line on the RHS will be considered a back-row attack and X8/V8 are the most likely suggestions. But the actual contact point of the attack is slightly in front of the 3m line, and if that's where the click is, then the front-row player gets suggested as the attacker, leading to weird suggestions such as the front-row middle making a back-row X8 attack.