The spirit-based mana regen forumula (infight) should be:

2.5 * (0.001 + Spirit * sqrt(Int) * 0.003345) + 1170

where 1170 is the standard regen (5% of base mana) and the other part is half of the normal out-of-fight regen (due to meditation). With following values:
Int: 5k
Spirit: 2.3k
Heartsong-Uptime: 60% => 120 Spirit at avarage
PT-Uptime: 25% => 125 Int at avarage

you gain ~54 MP5 more with heartsong (only spirit-based regen). Pre 4.0.6 heartsong will provide more mp5 (without AT since many paladins are not bloodelf ^^) then PT but with 4.0.6 and every DP synched to PT-procs, PT will give you more regen.

Not sure if that testing will hold up when actually used in game though. The regen from PT relies on DP being used for it's full duration and using DP in synch with PT which is likely not to be on cooldown. If you are in situation where you can use DP whenever you please then using Heartsong along with DP/AT on cooldown rather than trying to synch things would likely make a much better case for Heartsong. The other case that may come up is if you are in situation where you can use DP whenever you please the fight doesn't require any serious healing and what enchant you are using is irrelevant.

Would need to use a sim I think to get decent testing on it though.

Yeah, my numbers for Power Torrent are a bit of a best-case scenario and assumes fairly free use of divine plea, which again goes into the discussion of whether you'd like to cast divine plea and offset it with AW (and the increased 500 spellpower from PT) if you cannot afford the mortal strike debuff or if you want to save your AW and only cast divine plea in periods of light healing and use it less.

'Average' case analysis:

If we take PT to have 45s ICD with 12s duration, and DP/PT CDs are randomly distributed, there's only 3 seconds where you can DP freely without waiting for the proc (otherwise, PT will wear off before DP does). So 3s out of 45 you don't delay DP; every other moment you delay until the next proc (assuming you are prioritizing full use of DP/PT). With this usage profile, the mean wait time for DP is

sum(1..42)/45s ~=20s

which implies an effective DP CD of 140s instead of 120s. At this figure this is actually a loss compared to using DP on CD with no PT synchronization. This actually means the strategy is poor, however, rather than suggesting PT is a loss compared to no PT (which is silly). A better strategy is then not to look for complete synchronization, which implies the gain will necessarily be worse than the best case scenario. Waiting should only be preferable when you delay the DP by less than 8% of the CD (more precisely, the added benefit of having PT up), or about 10 seconds. This implies:

- 3/45 of the time you get full benefit from PT sync
- 10/45 of the time you wait a mean of 5s to DP
- 32/45 of the time you DP without sync

which is to say, the benefit is rather smaller than the ideal case (roughly 20% of the ideal case).

However, the 'average' case is not necessarily a good analysis since the DP CD and the PT ICD align pretty cleanly. In particular, 4 PT ICDs is equivalent to one DP CD. Therefore, the amount you actually delay DP (assuming you delay it) is however longer than the ICD it takes you to proc, times 4. Going by the above analysis, if you delay DP by more than 10s, it ends up being a loss; this means you need the average time after ICD for PT proc to be less than 2.5s in order to be a benefit above simply DP on CD. Alternatively, you could use it earlier and just hope that the proc comes somewhere before the last tick.

If you cannot DP freely, then aligning PT should quite difficult, and simply analyzing it as 25% uptime may be a decent estimate. Note however that for those who don't heal during DP, this is still optimistic since you are avoiding chances to proc within the DP itself.

tl;dr: Actual benefit from PT for DP is likely only on the order of 25% of the ideal case.

Edit: Er, yeah, 4x 45s = 3 minutes, so it'd be at 2 procs + 30s until the next proc, plus whatever delay. This probably is actually less advantageous to PT sync, however, since it's quite likely that you're just ending your 2nd proc when DP comes back off cooldown. Aligning it in any case in that scenario would imply lower uptime for PT anyway, so you'd lose the benefit from the other forms of regen.

Wouldn't it be simpler just to use DP every third PT proc? You would only be extending the cooldown by a theoretical 15 seconds every 2 minutes (more realistically would be 30 seconds, due to proc CHANCE and not guaranteed when its off of ICD). DP is a healing loss, so you can't necessarily say that using plea on cooldown is the most mana efficient method. Think of it this way:
0 seconds: You use HS+LoD to stack Darkmoon Card/Jar/Conviction, it also procs Power Torrent
5 seconds: Tank pulls
50 seconds: You're at 75% mana, PT procs, so you use DP (with synapse springs if engineer)
Assuming proc intervals of 50 seconds from this point on, the next DP/Torrent combo would be at:
200 seconds: PT+DP(+Synapse Springs, if eligible)

I know not many paladins are engineers, but even without that perk this type of setup creates less mortal strike uptime for relatively similar gain to using DP on cooldown. Basically all you're doing is waiting like 20-30 seconds every time DP is up to get an increased effect. Also, the int boost would help to offset the ms slightly (kinda like Talisman of Resurgence in wrath).

Edit: On second thought, increasing AT's cooldown by waiting would be a mana loss, so it's irrelevant to actual DP+PT usage. Might be similar for springs as well, since the nerf is making it so that it's pretty much only better than other professions if used on cooldown. 2 uses within 150 seconds wouldn't be an ideal use of springs, so that may not be relevant as well. Basically, my suggestion just DP after 2:20-2:30 intervals instead of 2:00 intervals

You can, but the point is that that's a loss (in terms of mana) compared to simply using DP on CD, in the infinite case (2m DP vs 2.5m DP is a 25% loss for 2.5m DP, which is only partially offset - about 8% - by the PT bonus). It's true that there are many reasons why you might not want to use DP on CD (most of which are encounter dependent) but the other part of my argument is that if you can arbitrarily time DP usage with proc, that implies the circumstances are probably such that you can use DP whenever it's up. In other words, if you can wait for the next proc to use DP without affecting your healing, why can't you use DP immediately?

In most encounters you will likely want to time DP with a foreseeable break, in any case, which is generally hard to time with PT.

The other issue is that by arranging PT with DP, you are also reducing the spellpower benefit from the proc.

I'm not sure, but the numbers I'm seeing seem a little off. I ran my own math using the formula given under Stats>Spirit and the base values of 5000 Int and 2300 Spirit. What I fond was:

Heartsong with 120 average spirit gain= 74.51 additional mp5
Power Torrent with 125 average gain= 17.74 additional mp5

The calculation for Power Torrent of course benefiting from Kings/plate.
As for any gain form DP/AT; you can't simply add that value without subtracting the mana lost form not using them on CD, as was stated previously. So the only real factors you can use are the ones you can depend on being there, ie. spirit based regen and replenishment. Though I didn't add the replenishment; these factors don't seem to over power Heartsong. Additionally with how spirit based regen works every new point of Int has a lessening value, whereas spirit remains constant. Add this to the fact that as your Int increases with better gear, the percentage gained from TP will decrease giving you proportionally less spell power. Meaning that, with no great changes to mechanics, Heartsong will gain strength with better gear and TP with lose it.

I'm not sure if my numbers are missing something, but I'm fairly certain my theory is correct. So it seems to me that unless you find your self reforging all of your spirit for another stat, Heartsong is by far the best healer enchant.

As your post hints, there is value to be found in a proc that is always beneficial versus a proc whose benefits vary depending on when it triggers. Granted, you can gear your choice of heals to better suit your new spellpower when PT is up, which in effect is an mp5 benefit, but it can also be expected that you will overheal more. Ideally PT will pop during periods of heavy healing. Realistically, PT will pop at both times when it's helpful and times when it's not. As long as you're under 100% mana, Heartsong's proc, by comparison, will always be helpful. Count in the mana lost from waiting to use AT/DP and I don't think you can make a decent mp5 case for PT.

More mana, of course, means more DL/FoL casts when they're needed. An oomadin is a failadin.

Hopefully those on the fence about the pure mathematics of the debate will gain some perspective from this.

In reality Power Torrent will be probably very unreliable. We would need to time perfectly three factors. PT proc, Divine Plea CD AND the boss encounter mechanics itselfs (we can't MS ourselves when the tank is about to die and we hardly try to keep him alive by spamming DL/FoL with use of some CDs). In addition, condition must be meet. Our mana needs to be under 100% (well, it use to be during the boss encounter, so we can avoid this factor) to get a benefit from proc (as a mana regen tool).
Anyway, it's easy then to assume how hard it will be to get the same effciency from PT and Heartsong. I think that we all know it. Personaly, I have a hard time sometimes when I try to time my Plea and Concentration Potion during the raid eoncounter, and I need to cancel Plea to avoid death of the tank or other players. With PT it will be a lot harder. It'll never be able to keep efficiency and safety of Heartsong.

I've tried out Power Torrent in last weekend's raids, and it's worked out pretty well for me. I was able to sync divine plea up with it almost every time, and I don't think I really lose DP usage. The longest encounter was our 9:20 Nefarian attempt, where I had 4 divine pleas synced up. The thing about DP is that if you get an 8:00 encounter, you won't be using it 4 times since you generally don't plea for the first time until a bit into the encounter depending on the healing required. And you gain the # of possible divine pleas at plateaus, rather than as a smooth function of time - this is particularly true if there are fights with downtime where we'd time our divine plea for that anyway. For me, at least, I don't think I really 'maximize' divine plea usage in that I use the max possible DPs during any encounter for a variety of reasons. For any encounter up to 6:00 minutes-ish, I don't tend to use more than 2, and only start eking out 3 if the encounter goes beyond 7 minutes, but that's very YMMV. So for me at least, I haven't really noticed a decrease in divine plea usage when I'm now trying to sync it up with power torrent. I've also found it pretty rare that there's such a heavy period of intense healing that I can't afford the MS debuff at all, particularly if offset by the 500 int buff. If I absolutely cannot afford it, I'd rather plea and use AW with it.

I think for people deciding on which enchant to use, they can perhaps check their WoL parses to see how many divine pleas you use for each encounter vs encounter length, and what your general philosophy on acceptable times to use divine plea and whether you want to couple it with AW or not. All in all, PT is probably not too far behind heartsong for regen, but the throughput synergy with divine plea is something to consider.

January 24th PTR Notes Update:
- Holy Light, Divine Light, Flash of Light mana costs increased by 10%
- Aura Mastery no longer works with Crusader Aura
- Lay on Hands is no longer on the global cooldown.

Not particularly relevant to us, but:
- Eternal Glory: The chance to not consume Holy Power from this talent now also applies to not consuming the clearcasting effect generated by Divine Purpose.

The 10% mana cost increase isn't that big of a deal really. 700, and 600 mana respectively. As for holy light, it's only 200 more mana. I frequent arena a lot, and when I face mirror comps being (Holy paladin, Feral druid), I heal 30 minute matches with that skullbash debuff on me. It isn't noticeable to the point of breaking us, so I don't believe it will do much in PVE either.