I have been doing some dummy testing in order to check/confirm some assumptions about haste and Flurry. I would like it if the experienced theorycrafters could pick over my tests and conclusions.
This is on 4.0.1, I don't have beta access.
I have done all the tests with
this 2/31/3 spec and glyphs, in Berserker stance, wearing gear with 0% haste and dual wielding Rock Hammer 3.7 (identical) speed maces. (In case any of that is relevant.)
Conjecture #1:
Flurry haste speeds up swings in progress.
Test #1:
Walk up to the dummy and engage combat from melee range to desync weapons. Record the time between MELEE_SWING/MELEE_MISSED events in pairs of two. These are affected by latency but should average out to 1/2 swing. Manually hit a special in between the MH and OH hits as often as possible, and never between the OH and MH hits.
Expectation #1:
If Flurry haste applies to in-progress swings, the two values should remain averaged to 1/2 swing time. If Flurry haste is only applied to swings begun while the aura is active then the greater number of OH hits with flurry active caused by the specials should reduce the OH->MH time and increase the MH->OH time.
Observation #1:
The MH->OH and OH->MH times both remain averaged to 1/2 swing time.
Conclusion #1:
Proven. Flurry haste applies to swings in progress, the same as it did when dr_ALLCOM3 tested it originally.
Conjecture #2:
Simultaneous melee hits only consume 1 charge of Flurry.
Test #2:
Attack the dummy until Flurry procs and then back away/spin around with 3 Flurry charges active. Move back into combat to activate a simultaneous melee hit and observe the number of charges remaining. (This is much easier to do if you sync your weapons as you have 3.7 seconds to back off.)
Observation #2:
In every attempt two flurry charges were consumed.
Conclusion #2:
Falsified. Simultaneous melee hits consume 2 charges of Flurry, not 1.
As an addendum to #2, it would appear that in the synced weapons case, if it is the first of the two hits that procs flurry, the second hit will consume a charge. I didn't do enough testing definitively confirm this though.
Comments, confirmation, testing on beta, or explanations of why my tests are flawed greatly appreciated.