Maybe I'm misunderstanding how people are saying spell queueing for casts under the GCD works. What I'm understanding seems to be common knowledge here is that if you cast a wrath that is under 1s cast time, there is no spell queueing, you have to wait until the 1s GCD is finished for your client to send the cast command, making the effective cast time 1s + time for the cast command to get to the server + key spamming time. While a 1s wrath cast would still allow spell queueing, making the cast time 1s + key spamming time, meaning there's a penalty for dropping the wrath cast barely under 1s, rather than just no gain.

I'm fairly laggy, 200-300 ms usually, so from a theorycraft point of view that extra .1-.15 seconds on my wrath seems significant when I'm deciding between starfires and clipped wraths. If I'm understanding this totally wrong please just let me know.

As for observing it, even with my laggy connection that would mean picking up the difference between a 1s cast and a 1.1 second cast from straight up perception, no visible numbers. That doesn't seem so easy already, skirting the edge of impossible with any better connection.

Edit: I realized it was the 3.1 thread where wrath clipping was being discussed rather than this one. I suspect I'm just misunderstanding what people mean by clipping though so this is as good a thread as any to be set straight.

I've been raiding as a Moonkin during the entirety of WotLK so far but at Patchwerk, I'm still sitting at a measly 4800 DPS. Studying the WWS thread for 3 seconds already explained why, I'm not using IS at all. I had no idea that it would be beneficial to cast, even though I run the Glyphed version.

So my question is, how do I weave it into my rotation, do I keep it up during Eclipse, Heroism? Should I clip it or let it run out and refresh during a more convenient time?

Also, when is the best time to use a haste pot?

If you cast Insect Swarm directly after Eclipse ends and refresh it 15 seconds later, you'll have a fair uptime without getting in the way of Eclipse. If you get unlucky and it runs out while spamming Wrath to proc the next Eclipse, it should be okay to refresh Insect Swarm. Do not cast it during Eclipse. Potion of Speed should be used during an Eclipse and combined with a clicky trinket if you have one.

I don't believe that we in the Druid theorycraft community have done nearly enough to nail down the interaction between haste, latency and the GCD. I tried to find this kind of discussion in other places (affliction lock, shadow priest), but searches for GCD were unrewarding.

The general modeling for the time associated with a cast is

raw_cast_time = haste_function(tooltip_cast_time, haste_effects)
if(raw_cast_time >= GCD)
actual_time = raw_cast_time + latency1
else
actual_time = GCD + latency2

There is very good agreement on "haste_function()". Values for latency1 and latency2 seem to be personnal guesses, with the "guess" for latency1 ranging from 0.0 to 0.2s, even for a good connection. Various simulators use latency2 numbers ranging from zero, to 2*latency1.

It is not too surprising that if one theorycrafter says that Wrath takes 1.0s to cast, and another says it takes 1.4s to cast, that we won't reach a consensus on when we should cast the spell.

Better numbers should not be hard to get.

- Find a target dummy. Make sure you don't have any on-proc haste effects.
- Turn on your combat log
- Cast SF thirty one times as fast as you can.
- Take a drink
- Cast Wrath thirty one times as fast as you can.
- Take a drink
- Cast IS thirty one times as fast as you can.

Record your haste rating, and tooltip cast times for SF and Wrath. Go into your combat log and record the time you started casting the first, and thirty-first cast of each type. For SF and Wrath, figure out how many of the casts were under the effect of NG. Record how you were casting (single keypress eyeball'd with Blizz's castbar? continuous casting via a mousewheel scroll?). If you have a mod that reports latency (recommendations?) record its reading. Tell us your results.

Answers to questions like "When should I refresh MF during Eclipse?" are just guesses if we don't know the ratio between MF cast time and SF cast time.

Well, actually, you'd likely want to spec using all of the haste talents except NG. And you'd probably want a shaman dropping WoA for you. Getting 50% haste is pretty difficult if you don't get help like that. You'd also want tests using all of the above plus Lust, and then making sure you have over 3.6% haste from gear, so you can test the clipping aspect under a steady state (at least for the duration of Lust).

For a GCD-length spell that doesn't invoke Nature's Grace, use Cyclone, Hibernate or Roots.

For a just-below GCD-length spell, Cyclone with PvP gloves might work (it depends on how they implemented the reduced cast time). I no longer have my BC gloves, and haven't gotten LK gloves yet.

In beta, Roywyn posted a script that will tell you the GCD for each spell:

/run hooksecurefunc("CooldownFrame_SetTimer", function(_, start, duration, enable) if start > 0 and enable > 0 then DEFAULT_CHAT_FRAME:AddMessage(duration) end end)

If you type that, it is supposed to tell you your cooldowns after each cast.

That script needs some kind of spam limitation in it.

Presumably what the game will tell you is the time spent casting the spell (or CD invoked by it) is not what's interesting here. Using that script just returns a spam of the official GCD enacted by the spell, which is pretty well understood (although it would solve the wrath or everything question). I used it in game with nourish then with wrath and it sure appears that NG only affects the GCD on wrath (or atleast doesn't affect nourish). That's on live.

Something more like what you describe initially is what I think would be best. Personally, I can't get non-NG wrath under 1s without heroism up. Maybe I'll be able to give this a try, maybe not.

My point for posting the script was just to determine if the PvP glove bonus works using haste, which would leave Cyclone at the GCD, or leaves the GCD alone while reducing the cast-time. If its the latter, it provides a convenient way to get timing tests on a cast that is just below the GCD.

I don't believe either t8 bonus should be difficult to model. It will increase the importance of correctly modelling instant casts, since we'll go from about one every 9s to perhaps one every 5 or 6s (depending on the IS->SF proc rate).

I think that balance druids were always meant to gear for a certain amount of crit rating. I was reading some of the blue posts on other forums and it seems that balance druids are meant to be based off of RNG now. Which makes for a very interesting mechanic for the class. It does feel a little sad to lose the 5% crit rating on Wrath and Starfire, but it looks like we will be gaining more in the long run.

I'm really not going to worry about this until semi-finalized gear stats are out, mainly because it's hard to tell exactly how much crit we're going to be working with when 3.1 hits now, considering the talent changes(imp scorch, imp SB in particular).

For all we know, full T8 might 5% more crit on it alone.

It's in WC as well (I'll release the next version once we know more about the 4p bonus), and T7-4 is worth ~165 DPS, T8-2 about 67. Granted, this is using my gear from a couple resets ago, without changing anything but the active set bonus, but there you go.