dazer
commented
13 years ago I included the response time in t12_4pc and updated the method a bit to have less nested 'if' logic.
The stat for matrix is set at the start of the cycle iteration in set_constants. I didn't test what happens when two of the stats have exactly the same value (I don't know what the game does in that case either). Ricket fireball is handled ala rocket barrage and included a method in stats to average it's damage value (I expect it to change somehow, based in AP damage or something).
As for Hungerer, I'm leaving it as is untill you write the logic behind the trigger. If the cycle is executed properly the trinket will proc eventually, but I suspected users would try to expend energy as soon as it comes off cd and that's why I modeled like so.
The approach I took for t12_2pc seems a bit naïve to me, but I couln't find anything better. I kind of reverse engineer the crit contribution directly from damage_breakdown. I thought at first to implement damage_breakdown as a dictionary of tuples (damage, crit component) but it kept breaking since each damage source is handled differently. Also, the outcome is a bit too low for what is expected reading the discussion thread in EJ.
As I commented on the issue for 4.2, here are the easiest to model new trinkets/set