The upgrade path beyond chest -> shoulders is extremely dependent on which other pieces of loot you receive, and the stats you gain/lose because of them. There isn't an easy way to tell you what the best path is until you are at the point of buying your third piece, you may need hit, in which case you would forgo the helm and pickup gloves/legs (can't say which is the off piece definitively until Lich King loot). If you are well above hit and need to drop some, you would obviously choose them helm as your 3rd piece.

The tricky upgrade path is if you end up with two pieces, about to buy your third, and you find out you are exactly, or almost exactly, at the hit cap. You would then need to rummage through the pieces of gear you have as off-pieces and tailor a new gear set-up which isn't drastically over or under hit.

Next would be the helm for sure seeing as the two last pieces, legs and gloves, are poorly itemized with hit and crit. Those are the two debatable pieces that you'd replace 1 or the other with an off-set piece.

If you are arcane and are hit-capped it will be best to go for the Chest+Hands T10. As you can get the legs from crafting this would be the best combination.

If you are hit capped you wouldn't want to go for hands since htey have hit on them. At the moment i'm seriously over hit cap (i'm speccing out of 3 pct hit in arcane and still over) due to some bad luck in drops and dkp item losses. Either way this means like a few others i will be going for Chest, shoulders, then helm supposing i don't miraculously lose 200 hit from items.

For any mages with a 258 chest + 4T9.245 (which I assume quite a few mages are due to the 2 excellent robes dropping from ToGC and the relative scarcity of 258 tier items), taking the T10 chest first without getting a token from ICC is almost certainly a dps loss and not as intuitive as upgrading from Merlin's or some other chestpiece. For us, taking shoulders and either head or gloves depending on if you need more or less hit rating as the first two pieces seems more logical if the goal is to get biggest possible upgrades. The transition is not as seamless, but without an upgrade token taking just 1 piece of T10 would be a downgrade.

Having 4PT9 245 + [Flowing Vestments of Ascent] a few days ago, when the time to choose my first T10 item came, I preferred going for the shoulderpad. Even if I loose the 4PT9 bonus 5% crit on Arcane Missile and Arcane Blast, it was still a better upgrade.
Besides, since the T9 and T10 shoulders have the same balance of stats, it seems a good choice.
Next one will probably be the helmet, because it seems easier to gain the lack of hit rating I would have then, than to loose the excedent I would have with another piece.

To add to what Maje said, you have to consider the marginal cost/benefit of swapping out a piece of gear. By itself the stats on the shoulders might look slightly better, but because you are losing the 4pt9 bonus, you have to factor in the 5% crit when you replace your first piece (but not for subsequent ones). For arcane, that's basically 45.9*5 = 229.5 crit rating that you are losing in addition to the old T9 shoulder stats.

This is why, as a general rule (note: I haven't done the math for T9->T10), most people would probably benefit most by replacing their offset piece (assuming that there is an offset piece at each tier level which has individually better stats than a corresponding set piece) with T(n+1) first, move on until they are 2T(n)/3T(n+1), then replace the last two T(n) pieces with a T(n+1) and offset.

I chose to take the 264 chest first, as said breaking the 4 set would have been a dps loss. However chest first then shoulders are both increases. Sucks to upgrade the 258 bit first over the 245's but this way you never lose out.

My plan was chest (got now) shoulders asap, then helm and gloves with the crafted pants.

Misread, please delete.

He clearly said the "264 chest". The 264 chest is perfectly itemized, with haste and crit. It is also 6 item levels above TOGC25 (half a tier). Thus it would be an upgrade even over 3.2 Best-in-slot: great itemization and higher item level means more intellect/stamina/spell power, too.

Regarding what tier pieces to pick first people should probably also consider that Toravon in VoA most likely will drop legs and hands. Depending on the rate you gather frost emblems in this may be a factor for you.

Question: Were you "spamming" (i.e. constantly mashing) your macro'd spells during the test in a shotgun approach, casting as one normally would (1 well-timed key-press per spell), or trying to precisely queue both its and your spells?

For a long time I have had my pet's cast waterbolt macro'd to my frostbolt, to ensure he is attacking my main target. If I am casting normally this works fine. But I am pretty sure that if I spam the macro, it actually forces my elemental to continuously restart his cast and never complete the bolt.

I'm not at home to test right now, but that's something to watch out for.

Spamming /petattack causes this behavior. Spamming /cast waterbolt does not cause this behavior. Your pet will not chain cast unless you issue /petattack at some point (unless you are spamming waterbolt commands). Once you have issued a /petattack, sporadic use of /cast waterbolt will not interfere with your pet's desire to chain cast. Don't throw out your keybinds for pet micro. But adding /cast to your spammed raid macros will reproduce Lhivera's results. Adding /petattack to your spammed raid macros certainly will not.

*delete*

I modified one of my combat log analyzer scripts to measure the average time between waterbolts. Ideally, one would see a below 2.5 second time because the elemental will gain from heroism (and possibly other raid-wide haste effects).

I took one log from December 23rd (ICC25, first 4 bosses) and another from Wednesday (ICC25, first 4 + Festergut + some Rotface pulls). The server was quite overloaded on Wednesday, causing us to actually fail on Gunship the first attempt (the guns would simply not fire fast enough). There is a noticeable difference between waterbolt hit frequencies, but it may be explained by raid composition (dec23: elemental shaman, jan7: demo warlock, so wrath of air was missing).

Dec 23: 2.5192
Jan 7: 2.6638

My network latency is typically about 70 ms.

I will probably add something like /cast [@pettarget,exists]Waterbolt to my frostbolt, fireball and deep freeze macros tonight.

1 - Yes, you create threat as normal after being given a -400k (if the complete mage compendium has the # correct) threat buff similar to the mechanics of fade, but threat is still generated as normal.

2 - The threat you create, post-modifiers, is the threat that is transferred.

I think the 400K is an important number...I believe the slimes on Rotface generate 500K threat on a single target correct? Which means that a mage under the Mirror Image's Buff would still get chased but the slimes as they form.

Actually, no. I unintentionally "tested" this last night. Every time I popped MI, I was inconveniently picked for the debuff. I would run to the OT, get cleansed, and the slime would go darting off. MI is a minimal DPS gain, so I've taken it out of my "I Win" macro for the sake of not causing a problem.

According to Wowwiki, its much more than 400k.
source

Unless things have changed, Mirror Image temporarily subtracts 410,065,408 threat (and then adds it back on 30 seconds later). Note that internal threat and threat displayed by Omen differ by a a factor of 100 so Omen would report this as a 4,100,654 reduction in threat I believe.

Details can be found in an old discussion at EJ about this here.

From the thread you linked, so unless I'm misreading what was said previously, both Kaytwo and I are wrong. It doesn't surprise me that I'm wrong, considering there are a number of other mistakes on the wowwiki section I sourced.

For simplicity sake, lets speak in Omen units. According to omen, the spawning add has 50k threat on whoever they spawn out of. Mirror image reports ~4.1 million threat reduced when you pop it, so thats more than enough to get the slime off you for 30 seconds, albeit this is stupid. If you're doing your job, you've kited the slime to the OT, and either A) you're waiting there for the next slime to spawn, so you can start forming a big slime, or B) you're kiting it into the big slime, and then getting back to your job.

After testing on Twins, while mirror image is active your threat is "masked". Your threat, and the benefit to the warrior, builds up until the effect of the images expires. On expiration your threat returns to the table, and the warrior gains the threat then.

This was tested by popping images on the pull, blowing all my CDs, and observing the threat given to the warrior. While the images were active he gained no threat from me, as soon as the effect expired his threat jumped up to match my contribution.