Wraht and what determines if it is -13, -14, -26 or -27 in eclipse gain:
It seems to be deterministic, every Wrath something like this happens
There is a counter which is tracked over all Wrath casts

1. Wrath is cast
2. Hit? => icrease the counter by 1
3. Euphoria? => increase the counter by 1
4. Is the counter >=3? => This Wrath will gain you 1 extra eclipse energy (-14 or -27)
5. counter mod 3

Here is the spreadsheet:
https://spreadsheets.google.com/ccc?...thkey=CJO0guQI
BarV is the value of the eclipsebar (-100 - +100) AFTER the Gain of the current row
Counter: Cells in this column are colored GREEN if wrath gains should be 14/27 and RED if they should be 13/26
+1/+2: how much the wrath of the current row will increase the counter
C%3: Counter modulo 3

Column E is colored depending on what the patter says
Column D is colored depending on the actual gain from the dataset

There are some weird behaviours:
around row 100
It actually should be
It seems like the two patterns
Can never be followed by -14 or -27, pretty weird.

Hamlet, wouldn't the simplest thing be to simply not cast IS before your first Eclipse? It's not a significant DPS loss since you enter the Eclipse quite quickly and it saves the frustration of accidentally activating HoI with 0 stacks.

From page 2

As someone just mentioned, this isn't worth the extra GCD. However, here's a little twist I haven't seen mentioned yet:

With Glyph of Starfire equipped, cast MF-IS-Starfire. The Starfire extends the MF and also applies NG to entire remaining portion. Compared to the usual IS-MF-Starfre, you get NG on your IS for free.

Maybe a moot point, since when you're using GoSF, your DoT's tend not to sync up anyway. But now that we know that GoSF extensions recompute haste on the remainder of the MF, it's something to keep in mind. It can work for you or against you.

Well, clipping IS on short movements is also an option, especially if using GoSF. With that setup, you're trying to maintain close to 100% Eclipse/NG MF uptime, and refreshing IS meshes much better. To be perfect though, you'd have to track your DoT's in detail to figure out which to cast each time you move.

As you said, I think the point really is moot, or at least half-moot. If the DoTs synced up properly, this would enable you to get NG applied to both IS and MF. However, in pratice you would never have a situation where you would be able to pull this off, except maybe opening a fight in Lunar.

If you tried to do this for Solar Eclipse, Sunfire wouldn't get extended until Lunar Eclipse, since you wouldn't be casting Starfire pre-Lunar. That'd leave the majority of the DoT without the added haste, and only a very small portion with it. It does actually work for the Lunar cycle though, because you are immediately extending Moonfire during these situations.

Btw, a small strange caveat... We tend to refer to our quarterly rotations as Pre-Solar, Solar, Pre-Lunar, and Lunar. That's a little counter-intuitive to me. Instead, perhaps we should start referring to them as Post-Solar and Post-Lunar. That way we could describe the term "Lunar Cycle" as the portion of the rotation beginning with Lunar Eclipse and lasting until Solar, and vice versa for "Solar Cycle." This way the language is more correctly in sync with the in-game functionality as it corresponds more closely to what nuke we are using. Or I could just be over thinking things again. In any event, that is what I meant by "Lunar Cycle," basically the portion of the cycle in which you are casting Starfire and using our Lunar (Arcane) spells.

A few posts back, Labrynth quoted Aceofsween from post 37, as showing the math for why IS->MF is better than MF->IS->MF. In post 38, MatsT suggested a possible error in that math, but I don't believe a math update was ever done.

Using the Tectonic setup from WC 110124 PTR (2t11, GoIS, GoMF, GoWr, 50% Dark Intent Uptime), under Solar:

IS tick: 5245
MFDD(0): 5379
MFDD(2): 9589
MFDoT: 6409

Instant: 1.207s, 11 ticks
InstantNG: 1.086s, 13 ticks

IS, MF
Time: 1.207+1.086 = 2.293s
Solar Damage: 11*5245+5379+13*6409 = 146391

MF,IS,MF (assuming second MF has LS(2))
Time: 1.207+1.086+1.086 = 3.379s
Damage: 5379+6409+13*5245+9589+13*6409 = 172879

Edits

With just this change, an Eclipse cycle takes 1.207s longer (1.086 here, and on average 1.207-1.086 later, due to loss of NG on some later cast), and does 26488 extra damage. The average DPS for the extra time is 26488/1.207 = 21.9k. The cycle as a whole previously had a DPS of 21.4k, so averaging this in would slightly increase the cycle's DPS (2.5% increase, 2% of the time, or about a 0.05% overall increase).

The difference is tiny. For this gear set, and during Solar Eclipse, it is not worth clipping a DoT, or extending the downtime of a DoT, just to allow both DoTs to be cast in a particular order.

The numbers seem a bit off, how does a 2 stack of lunar shower increase the moonfire DD to 9589 from 5379, and not to 10220?

Lunar Shower is additive with BotG and Moonfury. If you have both of those, it boosts the multiplier from 1.21 to 2.11, which is only about a 73% overall increase. Without BotG the increase is 1.15 to 2.05, or the 78% value you are seeing here.

I'm pretty sure you need to refresh your dots when there is only 1 tick remaining for it to be optimal, which means that you actually need to refresh them later than that. Our dots tick once every X seconds (with 1sec<X<2sec), so either you refresh them in the last second or you recast them right after they drop.
I'm not sure if clipping 1-2 tick too soon would be worse than waiting more than X seconds before reapplying, but there are other considerations like clipping 1 tick too soon in order to cast it while in eclipse , or waiting X or more seconds to reach an eclipse before you cast it..

Only the tick-rate seems to be recalculated, and nothing else. I tested it with both spellpower fluctuations and Eclipse changes.