I appreciate the responce, and all the work you guys do on this.
As for a solution to the problem raised in 2). I can think of a number of solutions though they varry in value based upon 1) the amount of additional processsing time that would be reasonable 2) the level of completeness desired.
Possible suggestions during build upgrade list (limiting the solution to weapon/off-hand calculations)
1) Recognizing that it really is only with one hand vs 2 handed weapons that this is a problem then I would suggest have Rawr check best currrent gear, if 2 handed weapon then in upgrade list itemize all off handed weapons and one handed weapons that are
'better' than the best one handed weapon/off hand available.
Of course this doesn't really get where we would like to be since there is no comparison between the current 2H and better 1H/OH but at least it is a work around. (BTW this is what I am currently doing to create a list for Lootplan)
2) Freeze the current gear metrics at their relative value, and compare every (possibly limited by at least one of each combination being marked available) 1H/OH combo vs 2H. Those combinations that are better than what is currently equipped would then be
listed. This of course introduced a known failing since i doesn't allow for on the fly comparisons and changes in gear metrics. But trying to do so would basically require a mainframe or a lot more time than most people could dedicate.
This I think would be the most practical options since while it would require more processing time it wouldn't be that bad.
3) Run a full optimizer on each weapon combination as part of the build upgrade process. This of course is impractical but would be the gold standard.