Well, Polearm Gamble states that the trigger is when an enemy enters an adjacent square. So the OA obviously occurs before he crosses the edge of that square.
Right, so entering one square is unsplittably linked with leaving another. That's perfectly obvious, on the face of it, but the point is that there may be other such linkages as well. The obvious question being: can you resolve damage (including resistances, vulnerabilities and temporary hitpoints) without actually taking any?
Compare with attack rolls: when an attack roll is interrupted, and things like shield resolve
before the hit, the effects of the attack roll aren't "saved" for later; you need to reevalutate the attack roll after the interrupt. You may say that the
attack roll wasn't changed by shield, but just the seperate (and subsequent) notion of
hit that involves comparing with AC, so that's OK; the
hit is the trigger - not the attack roll and just the
hit is negated. But many other powers don't work that way.
Consider Wizard's Escape: this power interrupts a hit and (RAI) teleports a wizard out of harms way. At that point, the wizard has clearly already been targeted and an attack roll made - yet once the wizard is gone (and presumably out of range), neither his defense nor the attack roll have changed (generally) - he is missed
since neither the target nor the attack roll is "remembered", and in the situation after the interrupt, the entire attack fails because the target is no longer valid. So,
RAI you can't
generally take the attack sequence apart and interpret individual steps as a temporally sequential listing of distinct points in time. Things that happen lower on the list can impact things that happen earlier on the list in
at least some cases; rather than a temporal sequence, it's just a bit accounting by the rules
. To take the accounting analogy further: if you transfer moner from account A to B this involves reducing A's value and depositing that amount in B - yet if depositing it in B fails, then no money is deducted from A. We speak of and evaluate separate phases to make things simpler to describe, not because they can succeed or fail individually. As far as I can tell this is the norm: you generally reevaluate the entire attack when the attack is interrupted; you don't split an attack phase-by-phase.
Now consider the orc:
If an orc leaves the reach of a swordblow before he takes the damage why would you "remember" the damage and still apply it even after he's left the range of the blow? Doesn't it make more
sense to reevalute the effects of that attack in light of the new situation? Doing that is also more consistent with other powers.
I still think you are creating your own definition of what constitutes an event here. The trigger defines the event. Resolve the interrupt before that event and then reevaluate the event. There is no rule saying that interrupts goes before an entire action. Actually, there are many examples that it doesn't.
I can't deny it: I am (to an extent) and
I'm not particularly happy about it either - I'd much prefer a clearcut rule. Unfortunately, the rules aren't explicit on this point. To be clear, I'm not suggesting that
actions can't be split, I'm suggesting individual attacks can't be split.
Lets take, for example, close or area attacks. These attacks presumably are "unsplittable" since you're
simultaneously attacking a bunch of creatures with a fireball (or whatever). Thus, it makes sense that even though they involve multiple attack rolls, such an attack counts as a single attack (as the rules explicitly state). But as soon as you permit triggers to resove half-way such an attack, that abstraction breaks down: what happens if the first burnt orc bullrushes his friend out of the flames? Or what happens when you interrupt not on "dropping to 0 hit points" but on being damaged by an attack (there are oodles of powers with such triggers) or even just being hit? How bout reactions? If you can interrupt not just an attack in it's entirety, but its the individual phases, why not react to a particular phase - which in principle anyone can do via a readied action.
If you evaluate attacks sequentially, then such interrupts can affect who's included in the area of effect or how they are affected. E.g. if someone pushes the swordmage out of the burst, can the swordmage then trigger his aegis since an attack that doesn't include him hits an ally? After all, you're executing attack phases sequentially, and at the point in time at which the swordmage chooses to use his aegies he is indeed no longer a target of the attack since he is no longer within the area of effect.
The sequence of targetting is undefined, and the sequence of attack effects is hardly defined - so for example, its nowhere said nor suggested that there's an interruptible phase between resolving damage and taking it. Indeed, those phases aren't even
named to the best of my knowledge; splitting is no less "creating your own definition of what constitutes an event" than not splitting it (as e.g. for entering/leaving squares). I see attacks as
generally unsplittable and further see "dropping to 0 hitpoints" as synonymous with "taking sufficient damage to drop to to 0 hitpoints" and as such would permit an interrupt triggering off dropping to 0 hitpoints to resolve before the damage in general, and in the specific case of an attack, to resolve before the entire attack.