Yes, by giving more miss/dodge from defense, for example. It's an elegant fix, simple to implement, and while it still wouldn't make bears want the strength, parry, or block on offset tank items, it would push the inflection point where bear tanking ability noticeably falls short much further out, quite possibly past gear available in WOTLK.

Many seem to be ignoring the additional penalty of focusing on a single stat that already exists in the itemization formula. I'm trying to keep my posts as easy to understand as possible so I'll avoid posting complicated formulas (anyone can look them up if he wishes). I'll rather illustrate this through examples.

Take 2 green quality rings with random enchantments. Let's say the ilvl93 one (Smoky Quartz Ring) and the ilvl120 one (Amber Band). These rings come with a variety of random suffixes; some offer a single stat while others offer 2 or even 3 stats. For this example let's focus on the "... of Agility" and "... of the Tiger" versions.

Smoky Quartz Ring of Agility: +24 Agility
Amber Band of Agility: +32 Agility

Smoky Quartz Ring of the Tiger: +15 Strength, +15 Agility
Amber Band of the Tiger: +21 Strength, +21 Agility

At ilvl93 the item budget could fit 24 agility or 15 strength and 15 agility. In other words, instead of getting 30 stat points, you got 24 stat points if you chose to go after a single stat, assuming strength was not desirable. That's a 20% loss in stat points.

At ilvl120 it's 32 agility vs 21 str + 21 agi. Or 32 stat points vs 42 stat points. That's almost a 25% loss in stat points.

Or, if you prefer the other angle, upgrading the ilvl93 band of the tiger to the ilvl120 version, you jumped from 30 stat points to 42 stat points, that's a 40% increase.

If you had the "of Agility" version tho, the upgrade when switching from ilvl93 to ilvl120 would be 24 -> 32 or a 33% increase.

Basically the whole system (DR and itemization) is set up around the idea of forcing you to balance stats rather than focus on single ones. Which would be pretty nice if there was more than one choice available. It would be fine even if there wasn't more than one choice available assuming same rules applied to all. As it stands now, it feels plain wrong being forced down a path you know you don't like just because there is no other path available.

Something like armourBonus = w1*STR+w2*AGI+w3*STA+w4*INT+w5*SPI ? (Perhaps with w6*APbonus thrown in as well?)

It would surely be a unique mechanism, well suited for the "druids are the guys that scale in weird ways" mentality, overall it would certainly scale with iLvl of gear, and it ought not to be too hard to find weights that would not give too strange border examples, so I like the idea of it but I suspect it is too complex to write a decent tooltip for in a way that the average user will easily understand the implications of it and know how to gear to take advantage of it unless all the weights are equal, but all weights (W) being equal could have weird results: It seems likely that either core feral stats likely to be on armour would not contribute enough to give a decent armour progression with iLvl while the player wore the right gear (W too low) or it would contribute enough to give a decent level of scaling in AGI, STA (the two likely to be on feral gear) in which case the additional armour gained through buffs might provide too great a difference between e.g. tanking with a spirit buff or not.

Short version: Blizzard is probably better sticking with KISS solutions like they usually do.

I'd rather you went in the opposite direction really. Instead of posting opinions and comparing green items. Do some heavy math, show your work.

I wanted to see a comparison of an epic ilvl 200 warrior and druid, and then see the comparison again at ilvl 300, where you maintain the same ratios of stats on the gear and post the numbers you get. We might get a slight scewing of numbers, since the ilvl 200 leather has a (too) high amount of expertise already.

Assumptions
To speed things up, i've rounded a lot of corners.
-I've assumed that all items have balanced stat allocation, spread around 5 stats.
-When selecting items i've prioritised items with 5 stats over 4 stats (starting with the t7/10 set, then filling in holes)
-I've ignored pvp items
-I'm not including weapons or shields. Warriors will get a stat-advantage there, but that is caused by us getting idols instead of guns/bows/thrown with stats on it.
-I'm not including trinkets, since it rarely makes sense to "scale them with ilvl".
-I'm ignoring sockets and enchantments, but i think the gear i selected has about equal amount of sockets for both sets.
-I'm only focusing on avoidance here.
-I'm using the warrior-dodge DR numbers for calcing mod misschance, but it only changes .1% so should be ok
-Agi bonuses are modded for SotF and kings.

Math
Going from ilvl 200 to ilvl 300 gives up the following itembudget maths
Epic Ilvl 200 chest ( (200-26)/1.2 )-> 145 ItemValue ( ^1.7095) -> 4952.9 StatPoints ( /5 stats) -> 990.6 StatPoints per stat ( ^ (1/1.7095) ) -> 56.6 points per stat
Epic Ilvl 300 chest --> 89.1 points per stat

Since slotmods don't change per item level, we get that going from Ilvl 200->300 will give you a ~57.5% in stats for all slots.

As always the spreadsheet is available at:
http://spreadsheets.google.com/ccc?k...ET8M9KeuCXUMcw
I've made a seperate sheet for comparison, and put up 2 sheets for parry and dodge for plate-wearers.

Druid stats, ilvl 200
584 agi
76 dodge rating
25 defense

Warrior stats, ilvl 200
0 agi
269 dodge rating
233 parry rating
450 defense rating (need to gem further def to get crit immune)

Results
Druid
+dodge went from 16.4 -> 23.9
+miss went from 0.2 -> 0.3
Total gain +7.6% avoidance

Warrior
+dodge went from 9.8 -> 14.4
+parry went from 7.4 -> 10.7
+miss went from 3.7 -> 5.7
Total gain 9.9% avoidance

Conclusion
Druids will fall slightly behind on avoidance as we go through the raiding tiers in wrath, with a total of 2.3%.
If you look at base values after talents, druids are already ~3% behind with naked stats. We do however gain ~750 armor from the extra agility, which isn't factored into the current result.
While druids can gain a bonus headstart at ilvl due to not needing to gem for defense, this is not taken advantage of in the comparison.
Even as we lose this head-start, we still only end up ~6% behind on avoidance. In the end game, Warriors should be hovering around 50% avoidance, giving them a ~ +12% TTL bonus, though RNG based. If druids end up with 10% mitigation or HP than warriors, i think we would be above par.

I think a mix from increase Bear-modifier and agility based amor effected is a good way .
If we add strength to the based amor effect to , we come closer and we have 1 more stat for scaling .
Im not the theorycraft expert , but i hope here is one that could show this on a math example (eg. Bear-modifier 5,3) .

Lydara

This is a preliminary discussion, involving way too many assumptions to deserve hard math in my opinion, this is why I prefer to keep this on a more abstract level. Still, we could indeed come up with some numbers based on existing items that would indicate a trend.

Trying to keep it simple again, I'll build on your example of the ilvl200 chests for druids and warriors and compare them to the ilvl213 versions since we know the finalized versions of these items.

As you correctly calculated, the ilvl200 chest carries 4952.9 StatPoints. Similarly, the ilvl213 one carries ~5602.1 StatPoints. I believe we all agree that the most efficient way to distribute StatPoints is by dividing them equally among the various stats on the item. As per your correct calculations again, that means 56.6 points per stat (stat mods applicable ofc) at ilvl200. For ilvl213 the number would be ~60.8. Since all stats and ratings we care about here have a StatMod of 1, we can say that the sweet spot would be at 56-57 for every stat on the item at ilvl200 and 60-61 at ilvl213.

Let's compare with the actual values now on the t7 chests of druids and warriors (examining only the stats that affect avoidance as per your example).

Druid:
ilvl200 - 74 Agi = 1568.4 StatPoints out of 4952.9 (~31.7% of the item budget spent on agi)
ilvl213 - 85 Agi = 1987.7 Statpoints out of 5601.2 (~35.5% of the item budget spent on agi)

Warrior:
ilvl200 - 47 def rating = 721.9 Statpoints
+47 dodge rating = 721.9 Statpoints for a total of 1443.7 Statpoints out of 4952.9 spent on avoidance (~29.1%)
ilvl213 - 52 def + 52 dodge = 1716 Statpoints out of 5601.2 (~30.6%)

This simple and completely abstract napkin math is enough for now to indicate the trend I'm referring to. As the ilvl increases, Agility is occupying proportionally bigger chunks of the available item budget to keep providing comparable benefits.

I don't see a reason within the context of this argument to go on with calculations on exactly how much armor/hp/avoidance will a druid and a warrior/pala/dk end up having at ilvl200 and ilvl213. We don't get to handpick the stats on the items, they are provided as is; and especially tiered sets are designed to provide certain numbers of mitigation and avoidance as deemed necessary by the developers. As per the calculations above, on entry level gear a druid spends roughly 2.5% more of his item budget for avoidance than a warrior. Moving up just 1 tier, the druid now spends about 5% more of his item budget on avoidance.

When the changes are finalized we can do the math to figure out how to make the best use out of the new game mechanics. The argument for now is that agility will most likely hog druid item budget and I will leave it at that until 3.1 when we will have the finalized items and formulas to determine exactly how much is too much.

Hey guys, what if +defense, in addition to affecting miss / dodge, was altered to synergistically affect armor (ie. increase armor)?

Now trinkets, rings, neck pieces with defense would increase our avoidance and mitigration. Would it be hard to add it into the game? Perhaps as a talent with Thick Hide?

Block and parry could possible be converted in a similar way such that all tanking classes would equally benefit from the same pieces and scale appropriately.

This is a trend that is worth noting. However, it is not necessarily a bad thing. Agility itself has a wide variety of uses and is actually worth it for us to stack beyond an equal distribution point. Considering that bears need to share gear with both cats and rogues, agility is the most common stat. It is also good enough for all three to justify spending more budget on. We definitely don't want them making all the stats on items equal in distribution even if it is optimal by the math (even if for no other reason than it's boring).

So, I'd say spending more budget on agility is fine. The problem is if even with spending that extra budget, will we still be falling behind? That is what we need an extra stat to overcome.

I'd like to brainstorm for a moment on Cat dps modeling... WotLK is fast approaching, and I'm trying to get Rawr.Cat updated and ready for release before then.

As most of you know, Rawr models are required to be actual 'models', not simulators; they need to always produce the exact same result with the exact same input, and results need to change gradually as input changes gradually (unless there really is a significant jump at certain non-variable breakpoints). Beyond that, it needs to be as accurate as possible, while maintaining a reasonable level of performance.

So, here's what I've got so far. In 3.0, at 80, there are many potential rotations, and which is best can change based on a wide variety of factors (set bonuses, talents, target armor, raid buffs, glyphs, etc). The conceptual differences between these rotations are not that huge, though, and can be expressed pretty simply.

All rotations agree upon a few things that are essential:

• Keep Mangle Up (no action necessary if Trauma or another Mangler is there)
• Keep Savage Roar Up
• Keep Rake Up
• Use Tiger's Fury whenever it's up, and you have <40 energy.

From there, we have a few variables which change the content of the rotation (I use the word Rotation here quite loosely, none are a set rotation):

• Use Shred or Mangle for CPs? (If Shred, still Mangle enough to keep the debuff up, if needed)
• Use Rip?
• Use Ferocious Bite? (If combined with Use Rip, Rip gets higher priority, and FB will only be used with spare CPs)
• What CP of Savage Roar to use? (Not counting the first SR, which should always be used as soon as you have 1cp at the start of the fight. Also, may simplify this to 1cp or 5cp, no option for 2/3/4, since those are quite unlikely to produce highest DPS)

That gives 16 potential rotations. With that definition of a rotation in mind, we can then move on to actually calculating the DPS of the rotation. Here's how I propose to do that.

Total OoC procs per second: Used to calculate what % of attacks will be made with OoC, to act as a constant % energy cost reduction.
Total Energy Available: Simple to calculate, just total up 100 plus the energy regenned during the fight duration, plus the energy from however many TFs you can use, plus 250 for each time you can use Berserk in the fight.
From that, subtract out the energy required to keep SR/Mangle/Rake up 100% of the time.
The remaining energy should be used on combo point generation, and finishers.

If just using Rip, calculate the number of combo points generated per combo point generator swing, and from that, calculate the energy required to generate 5cp, add the energy required to successfully land Rip. From that, calculate how many Rips you can do with the remaining energy. If more rips than will fit into the available duration, replace some rips with more combo point generators

If just using FB, calculate the number of combo points generated per combo point generator swing, and from that, calculate the energy required to generate 5cp, add the energy required to successfully land FB. From that, calculate how many FBs you can do with the remaining energy.

If using both, calculate the number of combo points generated per combo point generator swing, and from that, calculate the energy required to generate 5cp, add the energy required to successfully land Rip. From that, calculate how many Rips you can do with the remaining energy. If more rips than will fit into the available duration, then cap it at the max number of Rips that will fit in the duration. Subtract the energy required to do that many rips+cpgenerators. Divide the remaining energy by the number of rips to find the amount of extra energy per rip cycle. Calculate how many CP of FB you can do with that much energy, capped at 5. Any remaining energy, just use on extra CP generators.

How does that all sound?

I'm really surprised that more people aren't talking about the Str -> Armour conversion idea.

AP -> Armour won't work due to high AP fluctuations during raid encounters. But Str really shouldn't fluctuate much if at all. Kings is the only significant buff I can think of that would affect your armour with this mechanic. You don't have to worry about losing Kings buff when other raid members are dying.

Increased Agi -> Armour would just make us even more dependant on agility. So stam/agi would become even more so the only stats we'd be worried about for mitigation. Also, we'd have a problem when DR starts to kick in right?

Someone mentioned a few pages back that Str -> Armour wouldn't work because we lack Str on set gear. These changes (post xpac) will represent some serious design changes for feral mitigation and itemization. Whichever way they choose to go I'm sure that rearranging the stats on the feral set gear will be on option on the table. It just doesn't make sense to constrain themselves to something that was based on the old feral model. Adding Str to the set gear wouldn't negatively affect dps significantly either now that Str and Agi are so much closer together in DPS value.

More mitigation stats. More choices for how to focus tanking gear. More choices for food buffs, enchants, elixirs. Plus, it brings us inline with the shield carrying tanks who also get a defensive benefit from Str. Combining the Str -> Armour conversion with a deep feral bear form armour bonus talent would be a way to keep the conversion number low enough to not become the new golden stat, if required. As long as they balance the value that Str gives to make it similar to Agi so we don't end up with another situation where one stat rules them all, I think Str sounds like the way to go. Am I missing something here? A scaling issue or something? Why can't Str be that extra mitigation stat that we're looking for?

No, rearranging the stats on feral gear is not an option, because they want rogues and ferals to share gear. And they can't just change it on the feral tier sets, because then all bears would be limited to just tier sets, and then we don't share gear with rogues like they want.

I realize this is more of a complication, but I think that the optimal cycle will always use the most expensive attack when getting an OoC proc, so a flat energy reduction isn't quite valid. It's still likely acceptable, but it doesn't match the actual cycle going forward.

You said this much more succinctly than what I was going to post. I'm curious though, most people are concerned with the variability of our mitigation if AP buffs are taken into account; why not only apply the AP -> Armor conversion to AP from gear (the way armor works now with the bear form multiplier)? Someone suggested that already and I thought it solved the biggest issue with such a conversion system. Given the numbers posted earlier to make this work, perhaps AP -> armor is only part of the solution. The rest may need to be made up by an additional %based armor multiplier (somewhere) in the talent tree, but I think we need a way to scale (defensively) off a generic offensive stat that's going to be prevalent on both rogue gear and druid tiered pieces -- and AP seems to be the best candidate.

Aye, that's one of the three holes in my method that I see. It's not perfect, but close enough for now, I hope. (Really limited on time)

The other two things are Berserk's energy being spread out instead of concentrated, so you can't re-rip that soon, when Berserk's up, and that Rip/SR/Rake/Mangles may end at the same time, and you have to either let one drop for 1sec, or overlap one by 1sec.

Long time rogue going druid for wotlk. There is a ton of good modeling going on over at the rogue forums. Instead of trying to re-invent the wheel I am sure there is a certain amount work/calculations already done that you could borrow from. Otherwise the ideas you have laid out sound solid.

For druids more so than rogues there is the potential for a larger discrepancy between theoretical dps and actual dps. There are more things to watch (ooc up, tiger's fury off cool down, 100% rake uptime without overwriting previous rake, etc etc) and the sloppier your rotation the bigger the impact will be on your output. It might be helpful to add in the ability to adjust a % modifier on your capability to perform each function.

Let's say that you know you typically don't use tiger's fury the second it's up, but you waste 30 sec here and there. Under tiger fury uptime you adjust down from 100% to 50%.

With so many moving parts this will allow you to adjust a number of different variables to see how big of an impact they make on your overall dps output, and what changes to your rotations you should make to compensate. It will also give you insight as to what areas of your dps are most important and should be prioritized in terms of your attention.

From a coding standpoint this should be fairly easy to implement. Just add a variable for each aspect in the equations, and have the variable be pulled in from some area of user input.

I originally modeled berserk similar to this also. However, I noticed that it ended up changing the overall cycle in ways that I did not think accurately reflected it. Having higher average energy per cycle means higher CP per cycle. Then I would try to figure out how many of those CP went toward Rip (say 4-5) and the rest go to FB. This isn't really how it works out though, especially during the transition period in gear before you can sustain both finishers (depending on crits/misses/spec and others). I can go into more details if you want.

In the end, I pulled Berserk out into its own special calculation that is added in separately. My basic reasoning is that Berserk just modifies a single cycle and increases the number of specials (and FBs) during that time. I reduced the energy gain to 230 to be conservative. From that 230 energy, I calculate what the total damage would be for X specials plus one FB. This is somewhat of an estimate though, any Rakes during that time could theoretically be supplied by the base energy rather than the "bonus" energy (might need some refinement). I also compare what would happen if you split those extra specials into two separate FBs and take whichever total is higher. Note, these FBs are in addition to any FB that would be used during the normal non-Berserk cycle. If Rawr is set up to calc the dps of a cycle as a function call, I would call it once with and once without berserk and then merge them together based on berserk's cooldown and cycle times.

I know my model does not account for debuffs falling off at the same time either - that one seems quite a toughy. It assumes 100% uptime for Mangle and Rake (if enabled) now. My other problem right now is having CP for SR tied to the cycle. So, it's always doing something like 2SR/5Rip instead of floating a 5SR between other finishers.

Btw, will you be assuming a certain spec in Rawr? I know it simplifies many calculations, but it's hard to assume any of the important talents in WotLK. I'm sure people will be interested in dps for bear specs as an example.

Note - I also added an option to select the uptime for Replenish (assuming it still gives energy). At 100% uptime that's about 0.4 energy per second. Of course, I don't expect to have a druid keeping Rejuv up on me full time during a boss. So it might not be worth taking the time to code since even 100% is only about a 80 dps improvement, but I was interested in finding out.

Since TF restores 60 energy and your max energy is 100, x< 40 , x + 60 < 100 which will allow your energy to stay under cap. You need to look at energy deficiency, not energy. Also do note that if you are in the middle of a GCD when TF comes up, there is a chance you will lose energy by capping if you pop TF too quickly.

Mijae, thanks, that's excellent. Based on your advice, I think I'll just count berserk as 230 energy worth of extra combo point generators and FBs.

TF restores 60 energy. You want to use it when you have <40 energy, so that it doesn't push you over 100, wasting any.

I tried something out in game. It has no calculation in it but I tried different cycles for 2.5 minutes and then noted the dps done. I was lvl 70, some ok DPS gear. I focused on how to DPS in raids, not solo.
The following strategy/cycle/decisions on what to do when might be a good starting point to calculate dps done. This gave me the most dps. If some change to this cycle/a different cycle will give you less dps, you can probably remove all other cycles that are based on the same change.

What I came up with:
Indeed rake is much dps. Do that any time.

On Mangle vs Shred. Shred does more damage try to use it any time. However with keeping rake and mangle up, finishing moves etc. There is only about 1 shred before you cast your finishing move. So no simple: Mangle up and "shred till 5 CP" anymore. This means using mangle or shred doesn't make that much difference anymore since the usage of shred is not that big part of the dps cycle anymore.

On Rip vs Ferocious Bite. Rip does much more damage. Maybe critting FB does more damage but the thing is whenever you use FB you loose the extra energy you have. You have to wait for energy to come back. That is not optimal.
When trying to maximize DPS I tried to use rip all the time. Wait till Rip is over and then renew, even if you have all 5 CP up. Just wait a while before it is over and you get all the ticks in. Meanwhile energy will restore, which will allow you after casting Rip to quickly do Rake and Mangle again.
In normal cycles FB will be used once in a while. For instance when you waited to use Rip for it to finish, you have maybe 90 energy. Use rip. Still a lot of energy left. With some help of Omen of Clarity, some nicely timed crits you can have 5 CP again in about 4 seconds. Then don't wait for Rip, just do ferocious bite. However speccing talents that benefit FB are not really worth it in my opinion.

One thing I could not test was Savage Roar. However I noted Savage roar has a duration of 34 seconds at 5 CP.
Tigers Fury has a CD of 30 seconds. This is something that can be used very good in combination.

For instance:
fight starts:

• Mangle, Rake, Shred (till 5CP). Use Savage Roar.
• Tigers Fury (you have 60 energy extra, cause you will probably be able to use it all), Mangle, rake, Shred (till 5 CP) and Rip. Because of the energy you will be able to apply Rip a.s.a.p.
• Next try to indeed keep Mangle and Rake up. Do Rip just like described above. Wait for it to finish preferably.
• When Savage roar is about to end use your 5 CP to renew Savage Roar.
• Rip is about to end and you have no CP anymore. Use Tigers fury and get up to 5 CP again and use Rip again.

This way Rip is up on the target much of the time and does not suffer much of a "time-out" because Savage roar is cast.

Berserk. Now this is a "luxury problem". Good to have but it messes up your carefully planned cycle. Next to Tigers fury it is a good thing to sort of "restore" energy. Using it at the beginning is nice (Get Savage roar and Rip up fast) but you are likely to overagro.
Best time to use it is probably just after you have "had to" use Ferocious Bite, as described above. You will be low on energy, Rip will expire soon and you have no CP. Wait until you have 40 Energy; Berserk, and get your CP up to 5 a.s.a.p, Rip and continue the cycle.

One downside to the strategy of waiting for rip to end in this post is the following; while waiting mangle or rake will probably expire. I have no info on what is more hurtful. Mangle missing for 2 seconds, rake missing for 2 seconds or rip missing for 2 seconds.
I think Rake missing for a few seconds should be allowed. As it is a dot, the last tick should be allowed to do damage before renewing Rake. Thus my opinion would be; when waiting for Rip, after reapplying Rip, if Mangle and Rake are both no longer active, renew Mangle first, then Rake.

One additional thing: Whenever "waiting" cast Ferocious bite on your target, even if it is still active for an other 15 seconds. No harm in that.

I think you missed the point. We're not trying to find what the best DPS is (yet anyway), we're trying to figure out how to calculate all *potentially* best cycles. Your analysis shows what's best for you, but that changes based on many factors (Glyphs, Talents, and target armor being the biggest ones). Also, some flaws I see: SR is huge. Gigantic. Testing with out it and trying to apply anything you learn to level 80 combat is useless. Next, FB using up extra energy is bad, so you only use it when you have 35 energy. Also, to maximize the value of Berserk, you want to use it when you have as high of energy as possible, not as low as. The more energy you start Berserk with, the more energy you get out of it.

Ok I get your point. The magic formula: Add all variables and get the best cycle for you. That is way over my head at this point. But if you have it I'll sure use it!
As I never really dpssed in a raid and joined in from time to time (bit off-raiding) I wonder how you will be possible to keep to a cycle. I had to adjust all the time and adjust my cycle depending on things that happened. Some basic rules I can try to stick to (keep mangle up, etc), but as to getting a real cycle with debuffs expiring, a chance factor on Omen of Clarity and a crit doubling your CP gained, I don't really see any way you can really make a cycle. At least not like hunters are able to.
In my opinion, the only valid thing you can really do is find out what ability does what kind of damage, make a list of possible situations you'll find yourself in (Like much energy and rip and SW still active) and the best way to respond to that.
In that case you just need a calculator per ability on damage done. Though given CP generation pampared to the cost of energy is something to look to as well. (On that subject, Shred does more damage then Mangle as I said in my post. I forgot to mention Shred uses more energy, at least at lvl 70, and will thus generate less CP per energy resulting in less finishing moves.)
Correct me if I am wrong.

I do not totally agree. First I didn't say you should have "as low as" energy as possible. You should at least have 40.
Secondly; when you are at 100 energy it is possible you get Omen of Clarity proc and you would do an ability costing no energy. Then you would have wasted some part of berserk. Given 40 energy is a bit low on energy to use berserk, 100 is way too much. Better would be a value around 60 energy which should allow you to spam without much loss of berserk.

The more energy you start Berserk with, the more you can spend at double-value. True, getting an OoC proc near the beginning can mean wasting some energy, but the chances of that are low, and on average more than made up for by the double value of that extra energy. I brought that up because you said you use Berserk after a FB, which means you are at 10 energy.