That would be great, but I don't know how to find out which row I'm hovering over, so I wouldn't know where to move the button; the template shows no form of index.
I suppose I could just put in like a dozen invisible placeholder dummy buttons that become visible/invisible when hovered in/out and while the row text populates a separate hidden dummy button that each of those loot buttons would activate. I don't think that two superimposed layers can simultaneously have onhoverin triggered, but I guess if the button were small enough that there was a margin of the row item text all the way around it then you couldn't hit the button without triggering the surrounding text onhoverin to populate the action button.
It would convoluted and kludgy, but I think I might be able to make it work, and it would be less prone to mistakes at runtime, which would be a plus.
Though, if you know how to tell spatially which item just got hovered over, that'd be way better
(I'll definitely be playing with the code this weekend)