I've done some additional Hurricane testing and can probably confirm it has a proc rate of 1 PPM.

For my first test, I dual wielded 2.6s weapons with Hurricane on both. I had 787 procs in 17,267 landed swings. This gives a 95% confidence interval of .98 - 1.13 PPM. For the next test I used 1.8s/1.4s daggers. This gives an average swing time of 2*1.8*1.4/(1.8+1.4) = 1.575s. I had 264 procs in 10937 swings, which gives a 95% confidence interval of .82 - 1.04 PPM. Combining the data from both tests by normalizing by swing time, it gives an average PPM of 60*(787 + 264)/(10937*1.575 + 17267*2.6) = 1.015 PPM.

2.6s testing WoL: World of Logs - Real Time Raid Analysis
1.8/1.4s testing WoL: World of Logs - Real Time Raid Analysis

Edit after further testing:

Weapon based attacks proc Hurricane like a standard enchant. There's no ICD. It has a proc rate of 1 PPM. Each hand procs independently of the other, and the buffs stack.

The spell based procs are caused by a variety of abilities that wouldn't typically proc a weapon enchant, including but not necessarily limited to, deadly poison, rupture, garrote, and recuperate ticks. IP wasn't able to proc it as a spell or as a weapon-based hit. The proc behavior for these abilities is different from the weapon-based attacks. There is an ICD, and the shortest time I saw between procs when only using spells(LBs on a shaman) was 46s, making it likely it's actually 45s. The proc chance is probably 15% judging from this spell: Hurricane (DND). With Hurricane on both weapons, I saw 20 procs out of 74 casts outside the ICD(27%), which fits the 1-(1-.15)^2 = 27.75% proc chance you would expect if having the enchant on both hands gives it 2 chances to proc per cast.

As mentioned in a previous post, a spell-based proc when no instance of the buff is currently active will give you an instance of the buff. Normal MH or OH procs while this spell-procced buff is active will not refresh it and will instead start/refresh their own instances of the buff. This allows the possibility of having 3 stacks active at the same time. If either the MH or OH has an active buff when a spell-based proc occurs, that stack will just be refreshed.

If both the MH and OH have an active buff when a spell-based proc occurs, it will refresh one of them. Which hand's buff is refreshed seemed to be random. I thought it might be related to which hand had an active proc first or which had been refreshed most recently, but I saw conflicting results when testing both of those possibilities.

Lastly, when a spell-based proc refreshes what was a MH or OH buff, that buff instance is instead considered to be a spell-based buff. By that, I mean if that hand were to proc again, it would start a new stack of the buff rather than refreshing what was previously its own instance before the spell-based proc refreshed it.

Feel free to ask questions if any of that was unclear.

It seems a little odd that your control test had higher damage numbers than either of the other 3. If it was truly not affect Killing Spree damage then you'd expect the damage to be exactly the same across all 4.

What I did notice however was looking at the logs from the parses, in all but the very last trial, there were other people hitting on the same dummy you were hitting on. Any sort of debuffs they put up could cause variance in your numbers. I didn't see any debuffs put up, but I also didn't see Bandit's Guile in the log so clearly it's not showing everything.

Given the small sample size, the difference in damage between tests seems reasonable. With that small a sample size, variations of less than one percent from the mean as observed in Zephram's tests are not a significant difference.

I agree. I wasn't looking at his totals when I was commenting, I was actually referring to the raw data in the parses.

More concerning to me is
1) I thought Bandit's Guile was only proc'd by Sinister Strike or Revealing Strike? Neither of those abilities show up in the log.
2) As I said above, other people are attacking the same dummies in most of the logs.

Looking at the time stamps, it appears the log was edited and everything before and after the Killing spree was edited out. I say this because there's 2 minute gaps in the log and when Putok wasn't doing Killing Spree, apparently the other people stopped attacking as well.

This is relevant because of one other observation. He says he casts 25 Killing Spree's in each of the BG tests. And I counted the number of "casts Killing Spree" and it looks like he did indeed cast 25 on each test. However, Killing Spree "attacks every .5 seconds with both weapons until 5 assaults have been made and increasing damage done by 20% for the duration". So for every Killing Spree, there should have been 10 attacks. So with 25 Killing Spree, there should be 250 attacks.

BG 5% - 242 attacks
BG 10% - 234 attacks
BG 15% - 248 attacks

with no Killing Spree misses as far as I could see.

Looking at the log for the BG 10% since it had the fewest Killing Spree Attacks, I see it looks like sometimes there were only 8 attacks made instead of 10. (Note the 2 minute gaps at top and bottom. Those were already there, I didn't filter anything out)

[21:04:52.725] Putok hits Expert's Training Dummy 1 (O: 103)
[21:06:46.006] Expert's Training Dummy's Blade Twisting is refreshed by Putok
[21:06:46.275] Putok gains Killing Spree from Putok
[21:06:46.275] Putok casts Killing Spree
[21:06:46.275] Putok hits Expert's Training Dummy 1 (O: 115)
[21:06:46.358] Putok Killing Spree Expert's Training Dummy 1 (O: 186)
[21:06:46.794] Putok Killing Spree Expert's Training Dummy 1 (O: 170)
[21:06:46.794] Expert's Training Dummy's Blade Twisting is refreshed by Putok
[21:06:46.891] Putok hits Expert's Training Dummy 1 (O: 136)
[21:06:47.293] Putok Killing Spree Expert's Training Dummy 1 (O: 174)
[21:06:47.293] Putok Killing Spree Expert's Training Dummy 1 (O: 152)
[21:06:47.293] Expert's Training Dummy's Blade Twisting is refreshed by Putok
[21:06:47.760] Putok Killing Spree Expert's Training Dummy 1 (O: 188)
[21:06:47.760] Putok hits Expert's Training Dummy 1 (O: 134)
[21:06:48.144] Silvermoon Ranger begins to cast Shoot Bow
[21:06:48.144] Putok Killing Spree Expert's Training Dummy 1 (O: 154)
[21:06:48.144] Putok's Killing Spree fades
[21:06:48.227] Putok hits Expert's Training Dummy 1 (O: 128)
[21:06:48.445] Putok Killing Spree Expert's Training Dummy 1 (O: 197)
[21:06:48.445] Putok Killing Spree Expert's Training Dummy 1 (O: 155)
[21:06:48.445] Expert's Training Dummy's Blade Twisting is refreshed by Putok
[21:08:42.927] Putok hits Expert's Training Dummy 1 (O: 97)
[21:08:43.145] Expert's Training Dummy's Blade Twisting is refreshed by Putok


So the total damage done by this Killing Spree was 1,376.

I posted something like this a few days ago in the 4.0.1 thread Killing Spree Tests. I can add it here to this so that it's all in the same place if that would make a difference.

They were on a different dummy. I was on the level 60 dummy to eliminate dodges and misses.

The higher number on the control test isn't surprising given only 20 some odd tests, that's a pretty small sample, and its well within expected margin of error. It is interesting that there is a range on KsP attacks, rather than it using mean weapon damage.

The buff is not shown because i started/stopped logging between KsP to reduce clutter in the logs.

EDIT: Another round of tests. Complete logs, no start/stopping. Shorter trials because I'm a bit busy and tired, only did a control test and a 15% test and 16 trials. These were done on the level 60 dummy in SMC to eliminate dodge/miss on specials. Dalaran Sword is the weapon in both hands, no items with stats equipped.

Average numbers;
Control; MH hit: 182.8 MH Crit: 368.2 OH hit: 158.1 OH Crit: 319.8
15%; MH hit: 180.7 MH crit: 363.0 OH hit: 157.4 OH crit: 313.4

The disclaimer, again, is that the tests are too short to be 100% conclusive, but they show that it is extremely likely that something odd is going on. A GM communicated to me that the Dev team was "Aware and working on a fix" after my last petition, so hopefully that is the case.

Control parse: World of Logs - Real Time Raid Analysis 15% parse: World of Logs - Real Time Raid Analysis

I noticed a sort of interesting bug - Deadly Poison doesn't seem to benefit from the recent Assassination Mastery buff. With no mastery gear equipped and no talents specced, DP had a reported damage of 708; with Assassination specced up to Vile Poisons, the reported damage is 992 - roughly 40% more, as opposed to the 48% we'd expect. IP and VW both gain the full 48%, but Deadly Poison is only gaining 40%.

As an aside, pre-multipliers damage for the IP/DP/VW:

• Instant Poison - 352 + .09 * AP
• Deadly Poison - 540 + .14 * AP
• Venomous Wounds - 675 + .176 * AP

I didn't get a chance to do Wound Poison; if someone would like to test that, that would be helpful.

With a 1.4 dagger equipped MH with WP applied and a 1.8 thrown with no poison applied, I ran a sample of 1000 FoK hits and yielded 519 WP procs for a rate of 51.9% (50% expected). This would lean towards the belief that the poison proc cause by Vile Poisons is based off of the weapon it procs off of, and not off of the thrown weapon's speed. If you would like me to do a longer test, additional poisons, different weapon speeds, or any other combination please let me know.

Edit: To further my testing just to confirm or deny that for whatever ungodly reason blizzard would base the proc off of the MH speed only, I did another sample of 1000 FoK hits, this time with a 1.4 MH with IP, and a 1.8 OH with WP, and a 1.9 thrown with no poison. The results were: 661 WP procs for 66.1% (64.29% expected) and 327 IP procs for 32.7% (30% expected). While this goes along with the expected results of relative proc rate compared to the weapon it was applied to, I find it strange that over a 2000 sample the actual proc rate seems to be consistently around 2% higher than the expected proc rate, unless I am doing something wrong with figuring my expected proc rate or RNG was just kind to me tonight.

Given that those results are a good 8 standard deviations from the mean for a 1.8 speed weapon, I'm fine with saying that Vile Poisons FoK poisons use the speeds of the melee weapons for poison proc rate.

Edit: Regarding the edit: the standard deviation on a test of this size and proc rate is about 15 procs, meaning that you can be as much as 3% off before I even get worried. The fact that all 3 are off by a decent portion of that in the same direction is, admittedly, a little bit unusual, but I don't think we're into statistically worrisome territory. That said: I can't swear that there isn't some odd interaction or testing methodology glitch, and if people want to do larger tests to nail things down a little more firmly I certainly won't complain; but as of now I'm not particularly concerned.

I apologize if I missed someone else refute the bolded section, but I did some testing tonight that clearly goes against that. An excel of the cleaned-up log can be found here. The first two boxes represent hitting hemo once and allowing the buff expire. The third represents spamming hemo. As you can see, it treats reapplications as a refresh, and tics occur at three-second intervals from the original application.

My tests appear to confirm the remainder of the conclusions.

I just re-examined Ieatpaperbag's post. In comparison, it would appear during his tests, a new hemo strike would cause a dissipate effect, followed by a refresh effect. In my tests tonight there was no dissipate effect. Perhaps it was a bug that has been fixed?

EDIT: I'll add a log screenie before I hit the sack tonight.