With the recent changes to Eclipse, we see a good step forward. This model attempts to minimize the impact of movement on the rotation, allowing a Balance Druid to pick up where he left off with Eclipse still waiting to be used. Unfortunately, we trade one evil for another.

Issue #1

With the proposed system, we are guarenteed an average of 50% uptime with the Eclipse (+/- based on SS integration, but it does not scale). This is an unavoidable result from a system where the distance between procs is the same as the distance travelled under the proc. This can be simplified to a charge system, which I recall Blizzard saying they did not want to impliment. I'll use Starfire to elaborate, as it divides nicely.

Eclipse Power = Charges Let's let N be the number of casts in a Lunar Eclipse and C be the number of crits out of those N casts. Each Starfire generates +20 Eclipse Power. Each Starfire crit generates an additional +8 Eclipse Power. Now, let us suppose that a Lunar Eclipse Generates 25 charges. Each Starfire will consume 5 charges and each crit will consume an additional 2 charges. Thus, we have:

5N + 2C >= 25

Let C = 0 and we have N = 5 or 5 casts for the Eclipse to expire.

Let C = N and we have N = 4 or 4 casts for the Eclipse to expire.

Once this Eclipse has expired, I continue casting Starfire to build an additional 100 Eclipse power (to proc a Solar Eclipse), resulting in exactly the same equations and exactly the same average time. Granted, Starsurge has the potential to alter this equation, perhaps reducing the number of casts by 1, but that is a minor variation at best.

The same holds true for Wrath (only without the convinient divisors).

Crit Refund Model Let's suppose that each crit subtracted Eclipse Power instead of added to it while under the influence of an Eclipse proc. Then we would have

5N - 2C >= 25

Letting C = 0 and we have 5 casts for the Eclipse to expire. Instead of being an upper bound, this is now the lower bound.

Letting C = N and we have N = 9 or 9 casts for Eclipse to expire. This is a much larger range, allows Eclipse to scale with crit, and provides a mechanism to increase Eclipse uptime.

Minimum Time Model There is a second model that allows for increased Eclipse uptime. This model associates a minimum length for the Eclipse proc (I selected 15 seconds based on the current length). Allowing the Eclipse proc to last a minimum of 15 seconds or until the halfway point is reached (whichever happens [i]last[/i]) provides a scaling mechanism with haste and crit. Unfortunately, it also reintroduces some additional penalty for movement. On the plus side, it would never require more than 100 Eclipse Power to proc an Eclipse after the last Eclipse had faded.

Issue #2

Currently, once we proc Eclipse, we can still move the bar in both directions.  I do not see why we would want this functionality, as it encourages players to find whichever Eclipse is stronger and "milk" that Eclipse by extending the Eclipse to its duration and leaving the Balance Bar near that same Eclipse when the buff expires after 45 seconds. I can't speak for Blizzard, but I do not see this as the type of play they want to encourage as they already made a change to Eclipse to essentially force us to use both variations. I suspect that there will either be a 15 second cooldown triggered by the expiration of the Eclipse buff or they will implement a system where we can only proc one Eclipse variation after we procced the other.