Retroactively?
This is a very big assumption on my part and will obviously has to be tested, but if they go down the "animation based cd", then it'd seem logical to have a check of validity of effect at the end of the animation. And the mention of "completion of skill" makes me think that this is the case. So if you started a cast and it had a valid check (be it for flag or maybe for visibility), but then at the end the check was invalid - you got yourself a "failed" effect.
L2 worked like this. You couldn't cast abilities if your didn't have a visual on someone, but if you had the visual and then during the cast you lost it - the ability effect would not go through.
Anyways, like I said, it doesn't matter. If Hallowed Ground is the same, I have multiple tests to run myself on its flagging and corruption effects in Alpha-2. Until then, consider it irrelevant. I'm sure if somehow my plans for use of it are correct, it will be intentional design anyway.
I do think this matters, cause the same thing could be applied to aoe attack insteractions with the same switch and to even flagging itself. What if I do a forced attack on someone, but they move behind a tree during the cast? Do I get flagged because of my intention or do I not get flagged because the ability didn't attack (that is if the system I described is implemented).
But yes, it's all gonna be in the testing. I'll personally be rooting and giving feedback for the system to be like I'm used to, just because I see merit in it from my own experience and because I hate gcd with an absolute passion and it seems to me that animation-based cd allows you to not have a gcd cause it servers the same function.