I have been looking at options for a mates warrior, he kept telling me he would get errors opening old version of his gear in the new version and would have to reload from armory and start from scratch.
I am now getting an error trying to open the file I saved after loading it from armory yesterday with the 2.16 version, he is now logged out in his tank spec and gear now so I am stuck.

The following is the error it gives me, after I shut it down and restart it if I try and open the character again it opens but the lists show the Paladin and Deathknight pieces as well as the Warrior ones.

You're missing the point though. Sometimes I want to use my own item weights, and it seems like it would be pretty trivial to set that up in Rawr. Just make a "DPS model" which consists of (A*spellpower + B*crit + C*haste), etc., where all the coefficients are user-specified. Call it Model "Custom" or somesuch. This would let me make use of the powerful Optimizer functionality even when I want to tinker with some of my own numbers.

I'm still arguing with it even though I should be asleep.
I was getting all confused when my mate couldn't see the Paladin and Deathknight gear on his list until he told me he was still using the 2.2.16 version, for a while there I thought it was only my version screwing that up. He went to bed but I assume that it will do that for all warriors?

I just tried to create a new blank template to confirm this and got the following error:
Something isn't playing right.
OK. After restarting Rawr I set it to Human, Warrior, DPSWarr and straight away you can see Warrior, DK, Paly and Rogue hats showing up so it isn't filtering items by class.

Going back to the reason for this post, I was trying to set up a geming template using a [Nightmare Tear] and it just isn't showing up. I was interested to see if using one of them for activating a [Relentless Earthsiege Diamond] would be better than using two purple gems but I can't select it in a geming template.

I think the idea is that if you're replacing the model calculations the optimizer accepts with simple stat weights, it sort of defeats the purpose of the optimizer. I don't really see any advantage over using the various "loot rank" tools out there (on Wowhead, etc.) if all one is doing is scoring gear based on its stat "rankings"... with the possible exception of looking for the most efficient way to be just hit capped with those weights, or something. Still, that's such a simple task that it doesn't seem worth developing a Rawr model to me when (again) many websites already have tools to list gear based on stat weights. I'm guessing you're not going to find too many interested developers, leaving aside whether such a model is outside the Rawr "philosophy" or not.

No, Rawr could not support this, and I would be strongly opposed to anyone trying to make it do so.

Instead, you should ask yourself why you want to use your own weights, and solve the problem that causes that.



Looks like they messed with the built in filtering for DPSWarr, and screwed it up somehow. Will try to get them to fix it for the next version. This is just in DPSWarr, right?

Oops, delete me! Sorry.

While in the end I will defer to your expertise, I must also admit that after glancing at Rawr.Elemental I can't help but think that it would be pretty easy for me to hack that in the direction Arawethion was pointing.

However, it's your party, so you get to set the rules! I've certainly exercised my prerogative in SimulationCraft from time to time. And anyway, all joking aside, I agree with your position: Rawr.Custom is not worth the investment.

To be clear: While stat weights are regularly abused, I also believe that they can be effectively employed to make gearing decisions. Stat weights are merely the slopes of the DPS surface at a given point P1. Consider a point P2: How far from P1 can we place P2 before the slopes start to break down? If the gear choice is infinite then you are piss out of luck. If the gear choice is limited, even re-gemming may not move P2 far enough to make the "stale" stat weights produce a sub-par result.

This is the crux of the matter and why I believe you are correct to resist this direction. Development resource is finite. Just because something -can- be done does not mean it -should- be done.

There are two reasons (that I can think of) for wanting to use your own weights:

(1) You believe a particular Rawr module mechanic is flawed. But.... Why not invest the time making the module better and serve the entire community?

(2) You trust the Rawr module mechanics in the context of the scenario for which they were designed, but you wish to change the scenario slightly. For example, perhaps you are a caster who is suffering during movement fights, forced to frequently cancel casts and you want to weight Haste a little heavier. Perhaps you are having difficulty reliably burning down must-be-killed-now adds and want to weight Crit a little lighter to reduce RNG factors. etc etc

Rawr already provides optimization controls in the form of constraints (ie: keep my avoidance at X% etc). Providing stat-specific weights that work in concert with (not in place of) existing Rawr modules may allow users to perform custom gear optimization for their unique perspective. These weights would, of course, all default to 1.0. Increasing or decreasing these weights merely adjusts the signature of a gear set prior to passing it to the module for evaluation.

I well understand (and agree with) the root of your dislike for stat weights. They have been woefully abused, and the scale factors published in our SampleOutput wiki pages have perhaps perpetuated the problem even further. However..... Stat-weights in the form of "hints" may provide some useful functionality to an already exceptional tool without perverting its guiding principles.

My last words on the topic and then I'll let it go. Promise!

I think it would be a mistake to take the hints (1.6 in your example) and then try to force-feed them back in Rawr as absolutes. I would never ask for that. The weights I'm referring to would be relative..... and as such (reasonably) applicable to any gear set. Ideally, movement scenarios (and many many others) would all be supported by all the modules in Rawr. However, as much as I agree with #1 above, things are not always so absolute. Just because it is better to offer module-specific support does not mean a less-accurate cross-module solution is without value.

One thing that came up recently in a feral thread was the weighting of threat in the bear model. The threat calculation is based on TPS and then multiplied by the user's choice of factor to compare it alongside survival stats. During actual gameplay, TPS is important, yes--but what's most important is whether your threat gen is consistent or reliable. If two upgrades give me the same increase in my TPS, but one ups my expertise and the other ups my AP, I'll prefer the expertise upgrade every time. Even if the AP upgrade is a bigger upgrade in my sustained TPS, I'd still probably prefer the expertise. (Not having the threat you need in the next 3 seconds is more problematic than not having the threat you need over the course of a 5-minute fight.) The model doesn't make such a distinction, however; it's simply threat, multiplied by the importance you choose to give threat.

Now, Rawr does already have a way of coping with this; I can add a condition to the optimizer saying that I must be hit capped, and another saying I want my expertise to be at the dodge cap at minimum. However, what if I can't get quite to the dodge cap in reasonable gear for what I'm tanking? What if I don't mind being under the hit cap a bit, but would simply prefer to weight hit upgrades more than other TPS upgrades?

The next step is to suggest a refinement in the model, one that allows the user to compare both sustained TPS and threat... consistency, something, I'm not sure what to call it. Add a knob to that; allow the user to specify a higher weight for this reliability value, and consequently get recommendations that favor expertise and hit more heavily. In a way, you are adding a weight to hit and expertise when you do this; you're just doing so in a more sophisticated or meaningful way.

This is where I think the suggestion about a "cross-model solution" is interesting: Expertise does not just have this value for bears. All tank classes have to worry about parry streaks. Hit has a certain sort of value for everything from tanks to casters; outside of its direct contribution to TPS or DPS, it makes using any ability more reliable, smooths out cycles, and reduces the chance of human error. So, it's something to think about, anyway.

Burst threat is probably the term you're looking for.

Since we're talking about where modules fall short, any news on when we might expect ProtPaladin to be updated to current mechanics (i.e., Ardent Defender)?

Bug: Rawr is suggesting multiple Nightmare's Tears for me; sadly they are unique-equipped.