That's sort of a simcraft scaling factor question for your particular setup. Keep in mind that a lot of the benefit you'd gain from moving a few spellpower around would likely get lost in the statistical noise anyway.

Another option (one I'm partial to) is reforging for Mastery. If you keep filling your sockets with correct colors, you can reforge your over-cap stats into Mastery, which is often a decent stat as well, giving you the benefits of the socket bonus without the drawbacks. Of course, if (socket bonus + reforged mastery) < (10 int) then that's not a good choice.

I'm sorry if I missed this - did simulations confirm that Glyph of Imp > Glyph of Incin for Destruction spec?

Was testing this today, and it seems to me like the Imp's GCD is now scaling with your haste? Seemed to be sitting at 1 sec (or less than the cast time of a 1.4 sec firebolt) for me since 4.0.1a.

Edit: I didn't do any extensive testing and didn't properly take into account my lag (300ms), So I'm probably wrong in my above assumption.

For any claims as to pet/guardian cast times - can you please provide links to logs to verify the timings. Tooltips lie.

Tooltips don't lie, people just misinterpret them. The imp's Firebolt can be shortened under 1.5 sec, but his GCD cannot. Unfortunately, there is no tooltip that accurately shows GCD if you're past the cap.

I can confirm that the Imp is still GCD locked at 1.5 sec, as is the Succy.

Folks trying to min/max their dps will never reach the full potential of their gear by using /castsequence macros. Your dots and cooldowns have different timers, meaning that after the opening rotation, your cast rotation will almost never match your /castsequence macro. In addition attaching Meta to your filler spells, disregards the opportunities to stack your Meta with procs, bloodlust or decimate.

It's fine for one-button, mindless casual play, but if you're reading these forums on theorycrafting data to squeeze every bit of dps from your gear, you should really avoid /castsequence macros.

In addition, the reset condition in your macros, do not work like how most people try to use it. The reset timer is from the last time you called the macro. It is not a timer from the start of the rotation. This means your cooldowns will not actually match your macro reset timer conditions. see MACRO castsequence - WoWWiki - Your guide to the World of Warcraft.

I think the only /castsequence macro I used when I was demo (3.3.5), was an aoe trash macro, that chained meta, charge, immolate. As destro, I don't have any /castsequence macros.

I do use macros like:
#showtooltip
/cast [nomod] Soulburn
/cast Soul Fire
Which allow me to cast a SB:SF when Soulburn is off CD, or just SF when my imp procs.

And generic multi-purpose filler macros like:
#showtooltip [mod] Chaos Bolt; Incinerate
/use 10
/cast [mod] Chaos Bolt; Incinerate
/cast [@pettarget,exists] Firebolt

But in general macros which attempt to auto-select your spell are going to lower your dps.

"/cast [@pettarget,exists] Firebolt"
Seems to be useless.
Tested 10 min fight with dummy: clear fight and macros+autoclicker. Same amounts of firebolts both.
Conditions: I made a 1.6 sec firebolt cast.

Macros for preSB. Usable as SB button while combat and running. Usable as Harvest button as out of combat + standing:

#showtooltip Soulburn
/cast Soulburn
/cast Soul Harvest

Yeah, just copied from the post before mine. Macros has a notable animation on imp's FB. I'm testing on RU servers, so the macros differs a bit.

Tested once again:

1 try. Just the imp and the dummy, naked warlock 1 dark arts talent (tooltip shows 2.25 sec cast):
186(H)+50(M)+11(C)=247 firebolts cast
600sec/247=2.429 sec avg cast

2 try. With macros+autoclicker delay 300ms(3.33 hits/sec). Same parameters as 1 try:
192+41+15=248 firebolts cast
600sec/248=2.419 sec avg cast

+/- 1 cast error in calculations

Seems like same with/without macros ((
Ooooh and 2.429-2.25=0.179 cast delay is a huge loss. Sad.

Since your pets' spells can't be queued up like your own spells can, the effectiveness of the macro relies largely on the rapidity of your clicking. The macro will do nothing if your imp finishes a cast then starts casting again between two clicks of the macro. Your autoclicker leaves 0.3 sec between clicks, which isn't very helpful when your imp's pause is only 0.179 sec. i.e. the lower your latency, the more rapidly you must click to make the Firebolt macro effective. Personally, I spam my buttons about 4-5 times per second, and closer to 6-7 times per second when I really get aggressive, and I like to think (or hope) that most players spam at about 4-5 hits per second.

Here are my results from a few dummy fights:

Tooltip says 1.75 sec, 0/3 Dark Arts

	Fight Length	# of Casts	Cast+Pause	Pause Time
No Macro	194s	81	2.383s	0.633s
w/ Macro	193s	107	1.804s	0.054s

Tooltip says 1.51 sec, 3/3 Dark Arts

	Fight Length	# of Casts	Cast+Pause	Pause Time
No Macro	300s	187	1.600s	0.090s
w/ Macro	300s	190	1.579s	0.089s

Tooltip says 1.24 sec, 3/3 Dark Arts

	Fight Length	# of Casts	Cast+Pause	Pause Time	Pause After GCD
No Macro	194s	122	1.590s	0.350s	0.090s
w/ Macro	193s	121	1.595s	0.355s	0.095s

The first table shows a huge gap when not spamming the macro, which works out to about 39% of the imp's potential DPS being lost to lag. The second table shows an imp just short of the GCD cap, and the third is well past the GCD cap. Curiously, the latter two imps don't seem to be very much affected by lag while the first one is devastated by it. Does anyone know why this might be? Something to do with Dark Arts perhaps?

mistake in test?

I've been meaning to test for the possibility that this discrepency is caused by the firebolt being in flight causing the imp to pause before the next cast. Having the imp "move to" melee range of the target might allow him to chain-cast a bit faster. I'll have time later this week to follow up.