I am trying to decide whether to discontinue support for ShadowGreenLight at 3.1.

Reminder : before 3.1 crit added damage to dots exactly as if %crit was an added flat dmg modifier. Except that, contrary to how spellpower changes are handled, crit% changes are not applied on the fly to the dot tick values, instead we had to wait for the DoT to actually be hard refreshed. Coupled with the fact that SW:P is almost never hard refreshed this led to a situation where you could lose as much as 25% DPS on SW:P with typical raid buffs/debuffs (a net DPS loss of around 3-4%).

3.1 rolls in : dots can now crit. While I initially thought that meant that I could discontinue ShadowGreenLight I realized later that it depended on the actual implementation : what crit% value is used on the dice roll ? I see three main possibilities :
- when first casted (hard refresh) ?
- at dice roll time ?
- when refreshed by mind flay (soft refresh) ?

I'm not that good at statistics but I tried to find ways to reverse engineer that :
- I imagine that specifically testing for that in a controlled environment (ie for example with a balance druid turning 5%crit aura on and off) would require an insane sample size.
- Testing for that by pure datamining seems impossible to me.
- Last thing I could imagine was to have a small addon recording for each SWP tick if it was a crit, the crit% at tick time, the crit% when last hard refreshed, the crit% when soft refreshed... Then compare actual crit% with mean of each recorded flavour of crit%. What would be the sample size I would need to have a correct approximation ?

Is there something I overlooked ? A test that would be simpler to execute ?

I'm thinking it will be a soft refresh, otherwise everyone and their pet monkey will be popping Inner Focus>Shadow Word: Pain for a 25% boosted crit chance for the entire fight (That's a lot of extra DPS).

I do believe that you can cast a Shadow Word: Pain on a training dummy, and exit combat while the DOT is still ticking. This means you can strip naked, cast it, exit combat, put your gear back (use outfitter or something similar to do it quickly), and refresh it with a Mind Flay before it goes away.

By adding a full 20+% crit through gear, you should be able to more easily test if crit changes mid-fight affect your DOT.

Inner focus does not effect the damage of Shadow Word: Pain currently, and I doubt it would in the new patch either (unless it were bugged).

(edit: nevermind, not a useful test)

It currently doesn't effect the damage because it doesn't increase your crit chance, it "increases the critical effect chance of your next spell by 25% if it is capable of a critical effect," by the wording. If you tap it, it doesn't increase the critical effect on your paper doll, so it wouldn't effect your dots periodic damage.

It does not currently because dots can't currently crit (instead your crit gets added to baseline, like extra spellpower).

It should, under the new patch, since it increases critical strike chance.

Since shadow word: pain is not refreshed by recasting, the options would be (I might be forgetting something here):

1. Inner focus does not affect it
2. Inner focus affects it, but the crit is recalculated on mind flay refresh
3. Inner focus affects it, and the crit stays the entire time.

#1 is the case right now. #3 would be overpowered! I would most certainly abuse it. Wishful thinking, and feels unintended. As for #2... though possible, it means that there would be very little benefit if any to using inner focus with shadow word: pain, because a mind flay will be following it up in a few seconds anyway.

I suppose they could have the crit recalculate itself over some specific interval, but it does seem a little weird.

Last I checked, Recount could not track DoT crits.

I was testing earlier to make sure that the Moonkin Aura and the 5% Crit Debuff worked with DoT Crits (They do).

What I ended up doing is combatlogging, and then exporting the log to excel and filtering out/deleting all of the log entries I was not interested in. One of the lines in the combat log returns a 1 if the event was a Crit, otherwise it returns nil.

r1020 of Recount added tracking for DoT crits for the 3.1 PTR.

I ended up doing the same thing as Jdgaynor with regard to combatlog tracking. One of the things I noticed, and if someone has seen different then by all means let me know, is that at NO point did I have multiple DoT's critting simultaneously. This is something that really needs to be looked into, this is not limited to patchwerk style fights either. Think about high mobility fights..lets use Heigan as our sample.

during the dance phase I play the GCD game in between slime ruptures. it goes a little bit like this: while running I cast DP, by the time I stop moving GCD clears and I can cast VT, start running while casting SW:P. This allows my DoT's to tick at or about the same time for all three. If they are incapable of critting at the same time, I will be doing LESS equivalent DPS then pre-3.1 because now my DoT's not only are not able to benefit from my ability to crit, but also I no longer have the overall damage increase from crit chance I had with with pre-patch shadowform.


As for the question of DoT crit calculation: we could reasonably assume that our DoT's will calculate critical hit on a per tick basis. If it calculates our chance to crit for the duration of the spell on initial cast, it would likely do less damage over the duration. However. If the calculation for our crit chance over the duration of a DoT IS calculated at initial cast, I would see no reason for IF to not work for them. But honestly, this is more guess work than anything else.

I must have misunderstood.
Are you saying that if I get "tails" 1231231203712037120371230173013 times in a row with a coin, my next flip will get more than exactly 50% probability to be "heads"?
If that is what you are saying, I am quite certain that you are wrong.

DoT's criting will never ever modify the crit chance of an incoming Mind Blast.

If that is not what you meant, I am sorry, english is not my first language.