If you have an error in Rawr in my model, not a feature request but a real error or something that does not seem right or is calculated wrong, please post in the Issue tracker at Rawr, not here! That will get the issue fixed quickly, because if you post that here I will just tell you to do exactly this! Also, please note which version you are using.

Now that that's out of the way, onto the meat of the topic.

Rawr4 will soon be on it's way, replacing 2 and 3 in one fell swoop. I've already updated all of the calculations, and I've been able to add in a level 80 setting without much hassle. Some of the finer points such as double-dipping talents will are being worked on, but it should be quite, quite close to current and beta healing.

Current Model Progress:
In v4, changed burst-healing cost so it is reflected in Sustained's mana usage. All glyphs, talents, and spells (excluding healing rain) working correctly.


ToDo List:
Working on healing rain implementation. Also working on tying into the Boss handler on for more features.

While healing (especially reactive) is usually not represented by a fixed rotation (contrary to most cases of damage dealing), short rotations are very fine building blocks for your healing routines.
Percentages are fine, but they do not define cast sequences, while I believe, that having talents like Tidal Waves, CD-based abilities like Riptide, and setbonuses like 2pc T10 we just cannot ignore the synergy between our different tools, and moving from sequences to percentages would be a gross oversimplfication.

P.S. I'm not Antivyris, but I contributed some nice things to RestoSham model in the past and still looking at the way it develops, pestering Antivyris with "this doesn't work", etc.

I downloaded the latest version of Rawr last night, and didn't get much of a chance to play around with it yet. I will say the gemming now looks MUCH more in line with what I would expect. I will spend more time on it this evening and try to find some more constructive feedback.

Thanks!

This is actually how the original model worked before I took over. So far, from research and general gameplay, there are two types of us out there. Ones that abuse that riptide button, and ones that don't. I feel that, while they are only general rotations, they get the point across. Something such as the spreadsheet or the HEP report will almost always be more accurate simply because you can put in your exact style. However, that style may change fight to fight and night to night.

I think a good goal in mind, as far as Rawr goes, is getting pretty close to those two tools so you can go in with an expectation to how they will work. Like, if you are a chain heal spamming raid healer, it's good to know a decent expectation of how you would perform if you swapped some glyphs and went to the tank. My other major goal is if I can get this working close enough to the spreadsheet or the HEP report, it might be able to give some ball-parks as to where new trinkets or weapon procs lie as far as usefulness.

So, the short of it, while this sounds like an ideal idea, it's not really necessary to go that deep into the rotations. That said, however, there may be the options to select a custom rotation later, but I want the kinks worked out before that goes in.

I spent some time testing the new version and I'm glad to see that the results changed a lot. It's way more close to 'reality' now. I agree with most of Rawr's choices concerning equipment, except that Rawr doesn't go for 4 T10 but maybe it is because the 4 T10 bonus isn't fully implement into it yet ?
It tries to stop at 1268 haste rating, while the first reports of persons who are above the soft cap seems to show that haste is still the more efficient stat. But we'll prolly need more time to see aditional reports from people being over the 1268 cap to have more informations on the HEP it provides.
Rawr isn't always 'acurate' about gems choices : Rawr likes a lot int gems once you reach the haste soft cap while I don't really share this opinion. But well that's only a minor thing.

Thanks a lot Antivyris for the work you did on the fights' modeling which bring us a far more usefull and accurate tool than Rawr was until now. There are still a few choices I would do differently than Rawr does, but from the begining it's a tool designed to help us to think, not to think instead of us.

@Migosha
Rawr's preference of int gems does not come from reaching haste cap, but from estimated mana usage. If your mana-used-per-second is larger than mana-available-per-second, int/mp5/crit will start climbing up and up on priority list.

When I added support for Innervates (had to do that, since I run 10s and 25s with a loyal kitty who takes care of my mana, so I gear with a Vate or two in mind), I really should add the "Ignore mana issues altogether" checkbox for times where you want pure throughput modelling only, but's that up to Antivyris.

@Antivyris:
I don't know what you're working on, but I would like to suggest (and probably get involved) into either:

1. completing support for overhealing -- now that data on 4pcT10 is available, the model might could be almost complete (RT tick overhealing is tricky, but current implementation would suffice I say).
This should include several presets, along with an UI to customize models (I understand that my "tank/raid/heavy raid/self" models are a gross oversimplification, especially to CH jump modelling, and this data isn't readily available from online log parses, in fact the only capable tool I know of is shaman_hep, but it's completely different story).
2. sequence construction -- customized sequences, detailed existing sequences breakdown, constructing optimized sequences with respect to latency (while my attempts on latency modelling _look_ fine, they still do not properly take into account spell queuing, current model is hacky at least).
3. UI stats breakdown (tooltips on most of values like MAPS "you get X mana from gear, Y from Replenishment, Z from Mana Tide")
In other words, I would gladly continue fixing something here and there if you would indicate the preferred direction =)

After being in ICC10 a few times, I think I am starting to see where some discrepency is coming into play as far as mana goes, the amount of mana returned from popped orbs is actually a pretty decent amount that I think I need to get busy modeling in.

Also, the mana calculations are still a bit off, but they are a lot closer to what they were before. We are working to clean them up.

Update - Water-shield orbs popped per minute is now a choice in options that you can change. A lot of other things also cleaned up, as well as adding cast times on the main tab. Note, these are in the most recent commit, not the release version, so if you can't build the SVN you'll need to wait for the next release.

1. Set percents are obvious and are fine as an estimate, but they don't handle pure throughput changes very good (i.e. adding like twice more spellpower will push your overhealing over the limits), but any overhealing model is only an estimate, so I don't think this should get a lot of attention, just adding overhealing to healing target presets and/or the ability to input the overhealing values directly would suffice.
2. Latency ... yes, guess so if zero is the default setting, but I think that setting something around 50 by default might be a good idea (then again, maybe not).
3. Well, for me front panel is preferred. Tooltips are just fine and all in one place.

If at all possible, I'd like to get some feedback from Spreadsheet/Shaman_HEP users to see if there are any massive discrepancies between their numbers and Rawr's at this time. I believe the closer I can get to both the those, the better off we will all be.

With most of my logs, shaman_hep values haste highly, haste often is valued as 1.8-2.4:1 to spell power. On the other hand, Rawr (when MAPS>MUPS, otherwise it's entirely different case) weights haste over spell power only until around 1000 gear haste (w/both raid buffs enabled), with spell power really starting to take over around 1150 haste.

P.S. I don't think that Rawr is wrong in this case, actually. Stacking haste really works only up to a certain point (which is sensitive to nature of the fight, latency and ability/need to chaincast), at least for me. While for non-stop CH-only spam loads of haste (like 1500+) might be fine, just squeezing in RT on each cooldown or running around due to something bad going your way, devalues haste (due to 1-second gcd cap + non-casting latency).
It's more of a question of overheal scaling and the dreaded "haste saves lives argument". Even if haste == spell power in terms of raw hps, haste is ofter better than spell power due to increasing amount of potential targets, along with a possibility of saving someone reckless from an imminent death, yet I don't think this could (or should) be modelled at all.

TL/DR summary: Rawr is probably fine in terms of formulas, the only thing that needs checking is raw HPS from haste vs spell power at different levels of stats.

Currently a default is used for all the spells. These are averages that currently are pretty close to what I read scanning through a lot of different HEP reports.

Then, any of those averages for spells that can crit, an average is taken of those and then half of that is applied against critical healing to simulate the over-healing of crit.

Simple, I know, but it achieves very much the desired effect. This has been the safest route to take so far, because adding in 14 percentage boxes for putting in your own overhealing is a bit much, and just setting one single 'overheal' slider is not even close to accurate.

While Shaman_HEP is one of the best scales to gear off of, you do have to remember, it's only ever off what you did that night. That's why I'm trying to make sure Rawr hits the middle ground between Shaman_HEP and the currect spreadsheet by Diadalos. Striking the balance between Sim and Static is never easy.