This is very interesting because it suggests that for humans who have non-zero reaction times, camping would mean an even greater loss when compared to simply using FoF IL whenever possible.
There are other possible sequences where camping "munches" procs as well:
[#FoF charges=1]
Start casting Frostbolt
[Brain Freeze Procs during cast]
Finish casting Frostbolt, Cast FFB
[Frostbolt procs FoF, #FoF charges=2]
[FFB procs FoF which is discarded]
I believe Simcraft could not "cheat" in this case, because FFB has higher priority than 2-charge IL, but I haven't examined the log to find the actual causes of the munching. Even if there were no FoF proc losses though, charge-holding should only be something like <0.5% DPS increase, so the occasional proc loss could easily negate that.
After spending a few hours tinkering with frost APLs, I have some findings. Let me first state my final conclusion: FoF camping isn't worth it.
The default APL already includes a form of FoF camping, as quoted by Trelane. Whether it's some generic profile like durrty's, or the 4T16H sample profile, both do better with a "fire FoF IL as they come" APL. Obviously, if camping were worth it, there must be something that it's doing wrong. After talking with durrty, he suggested that we implement his strategies to avoid munching FoFs, which seems like the primary reason for any DPS loss. Here's a list of what we tried:
1. I stuck with the "simc instant usage of 2nd FoF proc cheat mode" for now. This means to stay with "fof.up" instead of "fof.react". If proc camping can't get ahead even with this cheat mode, there's no reason that a human would do better.
2. FoF munching due to Frozen Orb:
This one was easily fixed. Adding "action.frozen_orb.in_flight" as a conditional to stop camping and keep dumping FoF ice lances fixes this. Because frozen orb has to travel the default 40 yard distance, this more than did enough for getting rid of FoFs ahead of FO impact.
3. Dumping during meta procs to avoid proc saturation from haste LB ticks:
This one was also easily fixed. By setting a spell_haste conditional of >85%, "spell_haste>0.85", any frost mage not running a high haste build (sorry Akraen!) would be easily detected as under the effect of a meta gem or bloodlust. Using this conditional to dump procs meant avoiding some more cases where you might get saturated by procs.
4. Bomb snapshotting:
The logic behind this is that with bomb snapshotting of meta gem or bloodlust, the high haste would provide increase number of BF procs, which would lead to more gains from 2T16 applying to FoF Ice Lances. Bomb snapshotting is currently not implemented within the APL, but we did get some results. I'll post them in a follow-up post.
Combining all that stuff means that we get new conditionals for ice lances:
Code: Select all
actions+=/ice_lance,if=buff.frozen_thoughts.react&buff.fingers_of_frost.up
actions+=/ice_lance,if=buff.fingers_of_frost.up&(buff.fingers_of_frost.remains<2|(buff.fingers_of_frost.stack>1&cooldown.icy_veins.remains>2)|action.frozen_orb.in_flight|spell_haste>0.85)
The results were clear: despite very small improvements by adding these conditionals, none of these actually bridged the gap between camping and not camping. The loss of procs from camping ultimately outweigh whatever gains we could get from applying 2T16 to FoF IL instead of frostbolt. I tried the same with and without some of these added conditionals, but none of these really changed the conclusion. When you include human reaction times to Ice Lance usage, it only gets worse.
Considering added difficulty of camping, I can only say that camping strategies don't offer any benefits, at least in single target cases.
