Nobody who cares reads official WoW forums, EJ forums have turned to garbage some time ago and I can't start a new thread without being approved (and I don't really care), so I post here.
So I'm tired of arguing about Bloodlust and whether or not it affects global cooldown (gcd). I hate people who blindly think it does, offer no experience toward it, or they link an EJ forum post and treat it as gospel -- and then proceed to argue in favor of this misconception. In fact, EJ forum accepted defacto standards are STILL wrong in a LOT of cases.
I have had problems since forever ago with Bloodlust while searing pain tanking, so this is something that a warlock (who tanks or spams dots) or even a mage with arcane blast would notice fairly often.
I have no idea how anyone can even argue that it DOES affect global cooldown when nearly nothing else does, and proceed to provide zero evidence that it affects global cooldown. I don't think such evidence is possible to create, and that's why there isn't any.
First of all, if you still think Bloodlust affects gcd -- if you don't use a stopcasting macro, you are extremely unlikely to notice the effects of global cooldown at all, since your spells dont actually take 1.5 sec to cast.
Figure 1: The effects of typical WoW lag on casting.
I tested this in a secluded area with no people or npcs, 40ms round trip ping to the server (outside WoW), a high framerate, and a modified version of my combat log.
This is what happens when you don't use a stopcasting macro. You can't cast the next spell until you receive the UNIT_SPELLCAST_SUCCEEDED event (or possibly other events which occur at roughly the same time), which, for some reason, is happening ~250ms after the actual cast message hits. The are possibly other factors that delay your ability to chaincast spells.
So, if you don't use a stopcasting macro, you are likely not chaincasting 1.5 sec spells at all -- you are chaincasting 1.5 sec spells but 1.8-2.0 sec apart. Additionally, your finger spamming a button is probably going to be limited to a resolution of about 100ms (10 keypresses a second), so you won't really be able to always see the exact time the next spellcast is available to be cast.
Figure 2: Using a /stopcasting macro to bypass whatever server lag.
We see 1.579 seconds between the two heals, much lower than without the stopcasting macro, and much closer to the gcd than Figure 1. But this post is about Bloodlust, not stopcasting.
Figure 3: Bloodlust.
This figure is difficult to read because there are multiple things happening in parallel. With Bloodlust on, this rank of Healing Wave only takes 1.15 seconds to cast. I have successfully used stopcasting in this figure to start casting spell #2 prior to spell #1 going off. I use my experience to time when I should start casting spell #3. However, it fails.
Looking at the timestamps, it is clear that the failure message for spell #3 appears nearly 1.4 seconds after spell #2's cast sent message. So I timed casting between these two spells by 1.4 seconds, which is less than the gcd. The spell is only supposed to take 1.15 seconds to cast. Spell #3 never goes off. In fact repeatedly testing this, I consistently get failure messages (and spell #3 does not go off) if I press the cast key before 1.5 sec. If I wait slightly longer than the gcd (like 1.55 sec), then spell #3 goes off.
In fact, we can go beyond cast time spells and spam instant cast warlock dots. Popping Bloodlust, I've found, repeatedly, in good fps / low lag conditions, the lowest time between successfully cast dots is still around 1.55 seconds, which is still above the gcd.
* If Bloodlust affected gcd, then I should be able to spam this stuff every 1.5/1.3 = 1.154 seconds.
* Clearly, this is not possible.
* Therefore, Bloodlust does not affect gcd.
I challenge you to provide evidence to the contrary.
I don't care if it's supposed to and the client just doesn't let you for whatever reason.
The end effect is that even with Bloodlust up, you can't chaincast spells faster than 1.5 sec, and afaik, there's no way around this.