For me in full raid buffs, Rawr has the Glyph of AP at 57 dps and MA at 230 dps. It's not even close to being a decision which to be fair is expected, when you consider comparing a 3 second extension to an ability on a 2 minute CD, versus a flat % crit modifier for the entire duration of the fight. Even losing the T9 2 set is only a marginal change.

Just to eye some numbers: 3 seconds of 20% increased damage every 1.4 minutes is on average 3s/84s*0.2=0.71%

crit scaling is approximately a factor between 1.81 and 1.95 so lets assume you do 1.9x the amount of a non-crit.
With about 45% crit on average you would need to increase the crit by c, defined by
(1-(0.45+c)+(0.45+c)*1.9)=1.0071*((1-0.45)+0.45*1.9)
Solving for c leads to a whopping c=0.0111 i.e. slightly more than 1% crit. Any more crit you gain through the glyph increases your dps more.

The glyph converts 20% of your spirit into crit rating, of which you need 45.91 für 1%

0.94*45.91=0.2*sp -> sp=254.4 Any spirit above 255 will grant you additional crit, favouring the molten armor glyph.

Of course, if you have more crit, then the amount of spirit you need for break-even increases

with 60% crit you need 279 spirit
with 80% crit you need 316 spirit

I have about twice that amount of spirit in a raid, which would therefore require the gain by the 3 extra seconds of AP to be inflated by a factor of 2 just to be competetive. I highly doubt that you can inflate the gain by two just by cooldown stacking EACH time you use AP.

If I'm not mistaken, the cooldown on Arcane Power is 84 seconds (1.4 * 60) not 1 minute and 40 seconds. This would then increase the value of the AP glyph slightly:
Value of AP Glyph = 3s/84s*0.2~.71% increased damage

Looking through every post since 3.3 came out, is there an updated rotation table with dps, mps, etc as seen on post 1?

Please correct me if I am mistaken, but I'm pretty sure that if you start the channeling of arcane missiles, you invoke the GCD which puts a 1 second lock on any new spell. This means that you can't cut an AM that has just been initiated anyway, so the macro doesn't seem like it would do anything.

The only situation where I can imagine this macro making a difference is after casting say, evocation or blizzard, you decide to use a stashed away missile barrage proc from earlier arcane blasts. You'd lose some casting time due to latency. Of course, this only happens during high haste situations, since blizzard and evocation usually last too long for a missile barrage proc to be still around anyway.

As for the problem of starting a new AB right after MBAM and not gaining benefit from PtL, I am a bit confused as to why folks are even arguing about it. The effect of PtL is something along the lines of a 0.1 to 0.15 seconds shaved off an arcane blast, during a regular rotation. The time reduction is even less during haste cooldowns. The effect of using a [nochannelling] tag causes you to lose much more than that (0.3 seconds or more, based on my own experiences. May not be the same for others). Of course in an ideal situation we'd wish that we didn't lose any cast time at all, but since we can't fix server/client latencies anyway, why should we go out of our way and lose even more?

This is a nice point. The only AB that gains from the effect using a nochanneling macro is the first AB where you gain at most 0.2 secs off. That doesn't look like a big gain considering you are already losing twice your latency using a /nochanneling macro.

You also lose the last missile by clipping, which is worth quite a lot (half an AB at least, especially with full AB stacks).

I'm talking about the exact opposite, using a /nochanneling macro to not clip the AM and make sure the first AB after MBAM is hasted through 2pT10.

Correct, same result here

On the topic of glyph of AP: the new magegraf does account for cooldown stacking including bloodlust, 4-piece, icy veins and AP (but not a trinket, but this is a much smaller component). With all that in mind, over a 5 minute fight magegraf has glyph of AP as 34 dps less than glyph of MA. A lot more favorable than the rawr calcultion, but MA would still be better and more consistent damage.

Edit: Comparison done with best in slot gear.

I have updated theorycraft results for the new changes, posted here with BiS gear:

Vontre's Magegraf