My only guess is that crit damage buffs work a little differently. Like Storm Power, for example, does at least some live updating, so on my feral, when I lose the buff, my Rip tick crits are normal without re-applying and I can't cheese Storm Power by getting a Rip up at the last second. That's the only exception I've seen so far to the oldschool "buffs take effect on cast or dot application"

Measuring the crit chance of a 4th spell under the effect is rather complicated to say the least, and would take a lot of data to prove conclusively.


What I did was this:

Set up gear on a premade to have as high of a crit chance as I could, get naked, cast combustion, fireblast until a crit happens, scorch until another crit happens (applying Imp Scorch), put gear back on.

This gives me the highest chance for a high stack of combustion.

Then, Pyro->Fireblast combo. What I observed was a Pyro->Fireblast combo consistent with getting the damage bonus from combustion on both spells.


What spell combination are you using to produce the fourth spell? There may be a difference in instacast no missile and instacast with missile (Fireblast / HS Pyro)



Edit: So 10 datapoints using Pyro->Fireblast returned 9 events where both got the damage benefit, and 1 where the Pyroblast didn't. 10 datapoints using Fireball->PoM->Pyro to simulate HS pyros produced 2 events where both got the damage benefit, 2 that didn't get both as crits (but the other spell in both events got the damage benefit), and 6 where only one got the benefit, usually the fireball (so the non-instacast).

Reasonably convincing to me atleast that travel time definitely is involved.

So, what we're looking at is the following model now:
1) Crit chance is determined when finishing the cast. As are spell power and damage multipliers (Maulgar/Illidari Council tanking).
2) Crit multiplier is determined when the spell actually hits.
3) Fireball => Fire Blast will usually give you two crits with the higher Combustion damage, Fireball => Pyroblast will give you two crits with only one having the higher Combustion damage.

Now that's some awesome headache to plan around!

Is it really accurate to assume though that the effect of pom pyroing, which would be 2 gcd's rather then the direct one gcd of a HS pyro ,to mimic HS/pyro is valid? What if the effect of casting pom is causing you to not get a quick enough pyro to tail into the spell, like say if the multipliers and such are being added upon spell completion/spell landing. I was just wondering if this method could possibly be affecting the pyro getting the combustion benefits with the fireball. Based on his data he stated that 2 of the FB/pom+pyro's did get the multiplier. Just wondering how this is all working....

PoM isn't on the GCD. PoM+Pyro is instant.

I understand that, but on the PTR with no one else around buffing me and molten armor turned off (to make my crit rate as low as possible) I've got about 30% crit rate on my pyroblasts, and in over two dozen tests of fireball->pyro situations now with 7+ stacks of combustion the fireballs and pyros have *both* always crit. 0.3^25 is so astronomical as to mostly prove it to me. Just one non-crit would prove otherwise, but I haven't managed to find that anti-proof yet.

I tried fireball->HSpyro and fireball->fire blast. In the latter case, the fire blast benefited and the fireball didn't, in every single test. In the former, the fireball benefited and the pyroblast didn't. How it would show up in my combat log was the first crit, then Combustion fades, then the second crit, and the crit in the log after the Combustion fade never had the extra 50% bonus critical strike damage. In your tests where both spells benefited, did Combustion fades show up after both damage events or between the two? It could simply be a matter of lag in the buff fading sometimes allowing both to benefit, similar to other such effects like FoF ghost charges. I'm interested in seeing a sample where the buff fading shows up in the log before a crit with extra bonus damage, since I've yet to be able to mimic that case after now adding another couple of hours of trying.

My impression, based more on the way shatter combos and FoF ghost charges work than any tests with Combustion itself, since it's very hard to determine whether +10% crit is or isn't working, was that the fireball in the air when Combustion is first activated doesn't gain any crit chance. It was generally considered just a waste of a Combustion charge when that midair fireball crit. Crit chance is definitely calculated on cast for the purposes of Shatter, Fingers of Frost, and sneaking in a fourth Combustion "charge" at the end of Combustion (since with more Combustion stacks, it's much easier to determine that the extra crit chance is effective than with just one stack; +70% crit is pretty easy to test, especially when the final value is above 100% as in my previous post; +10% crit is very hard to test unless you can somehow get to a situation with no crit chance or 90% or higher crit chance pre-effect).

It is also easily tested that the damage effects of spellpower are calculated on cast instead of hit as well; just watch when Praxis procs on your 2pT8, and notice the spell that proc'd the effect doesn't gain any benefit from it, but following spells do. You can also /cancelaura Praxis while a spell is on the way to the target and it will still do the damage including Praxis rather than without it. Use trinkets will also prove this. However, it seems critical strike bonus damage specifically is calculated on hit instead of cast, unlike every other effect of which I'm aware.

Well so the new model we have is that basically this:
On spell cast
-> the 2 rolls are done, basically determining the outcome between miss/hit/crit.
-> the final hit damage is calculated, whether or not it crits (and presumably miss)

On spell landing (and not missing)
-> if the spell crits, the final damage is multiplied by the current crit multiplier.
-> presumably ignite is calculated/applied here. If my ignite theory still holds, I had concluded that debuff application time was related to the bug, and likely explains why ignite munching still exists.

color coding: untested

For what its worth, we could test crit multiplier being calculated on land and not on cast by spending a talent point in burnout before a spell lands. I mean its a lot of trouble but at least we could know if the behavior is (somehow?) specific to combustion or just the way things work.


edit
Actually, come to think of it, all of this makes perfect sense. The last theory I had posted about ignite munching fixes was specifically about why HS-pyro wasn't ignite munching all the time. My model was that roughly it would track the mid-air spells for calculating ignite damage. Actually, heres a link to the last I posted on the matter of how ignite functions [Elemental] 3.1 PTR changes and testing / http://elitistjerks.com/f75/t37364-f...9/#post1168006 . But the whole underlying point here is that there was no clear model as to why hs-pyro didn't produce the expected amount of ignite munching. If the ignite damage is calculated upon landing, then it means as I suspected that debuff application time is the sole cause of munching, and furthermore explains why ignite munching was much lower than we had expected.

This is interesting. I've Shatter-combo'd a lot to take advantage of what I always thought was the near simultaneous landing of two spells on a frozen target, but I was also under the impression that if a frozen target becomes unfrozen half way through a cast, I lose the Shatter effect on that cast.

I've just realized these two things don't make sense under the general assumption that all crit chance is calculated on cast. So either the second assumption here is wrong (that all casts that start on a frozen target but end up hitting an unfrozen target lose the crit chance increase), or the first assumption about Shatter is based on some other mechanic.

And just to confirm, yes the fact of a spell critting, or not, is determined on cast.
Just ask any boomkin, you can fire off a wrath and have natures grace proc as it leaves your hands. Though it wont actually trigger Eclipse until it lands on the target

Could this be tested with metanoia? Midair nuke + spellsteal.

deleted

CoE won't affect spells that were already in the air when CoE is applied, if that answers your question. I do remember testing that a while back to see if there was any gain with my warlock doing shadowbolt/CoE rather than spending a GCD casting the CoE before shadowbolt. Modifiers like that are easy to test with locks' deathcoil, since there's zero damage variability on the spell.

Ok, so I think there is a third option then that I missed. Crit is determined not on cast beginning, but cast end. That makes more sense and explains why you don't get the Shatter effect on a frozen mob you started casting on that is unfrozen by the time you finish.

It would still be nice to prove that current combustion cast in mid-air has no effect on crit chance, as it definitely counts as a charge.

SimulationCraft r3288 is available for download and contains the most recent ptr updates.

While the simulator can model travel-time mechanics, it is only done for a few spells. Once the TC on Combustion travel-time effects firms up a bit more I'll make the appropriate changes.

If i did understand clearly some of you tried to test if you could, by some bug get 4, not 3 crits from Combustion. What you should be aware of, first, is that, if the time between two spell hits is relatively short (lets say 100ms) then lag plays an important part in how the game behaves, that`s why combustion buffed both your fireball and fireblast but only fireball from fireball->HS-pyro combo

e.g.:
You have 130-150ms lag, which is not that evident, and can occur often.
You are on the last combustion crit and cast fireball, then fireblast your firebal will hit the target aprox. 0.2 secs after it leaves your hands. Fireblast will hit it after fireball leaves your hands with about 0.1 secs due to imperfect human reactions (if you don`t count lagg, but lagg will affect every spell equally so adding it only uselessly wastes our time here.) Due to the fact that the time between the hits, 0.1 secs is smaller than your lag, the game will not realise fireblast did crit (and remove combustion buff from your spells) until AFTER your fireball hits and crits aswell. But, then again when you Firebal->HS Pyro they are both missile-based spells, due to the fact that both have a traveling time and then your game has more time to figure out the crits and remove the buff. Due to the fact that your reactions vary (and do take time! about 0,15secs or more) the time from when fireball leaves your hands and pyroblast hits is reactions+travell time(+lag) whilst for fireblast its just reactions(+lag). This travel time of 0.2-0.3secs is enough for the computer to sometimes (depending on your reaction time and lag) realise the crit before fireball lands and modify its crit chance accordingly. Therefore a combination of Fireblast hitting quicker than Pyroblast and your faulty net connection are responsible for the variable 3 or 4 crits, not the game itself.

I'll make a wild guess and suppose that what makes people debate about it is that new mages aren't aware of the old mage TC, which had many times proven how spell mechanics work. Its gotten so ingrained that nobody takes the time to mention it anymore, although I could understand that a new alt would possibly not know since the information is pretty much assumed to be known by others. In any case, my previous post pretty much explains in precision how spell casting function, correct by our newly gained knowledge.

I have a question about the misery talent increasing spell damage by 5% on the first page listing all the optimal raid buffs. I cant find any spell tool tip suggesting this. Is that correct?

That whole post is really out of date (some of those things haven't existed since early in WotLK beta). You can safely ignore most if not all of it.

misery provides +3% hit chance on the mob now, not +damage.

Yeah thats what I was thinking. Was just confused when I was reading through that list. Thanks.

Just an observation, to anyone that's doing Twin Valkyrs as fire, LBing the second valkyr is a DPS loss if you're not tanking them close to each other.

The value of fireball on your target is fireball*1.5 and LB on the secondary target is LB*0.5, inputing the DPS value for each spell from rawr yields that expanding a GCD on off-target LB is a waste of DPS.