I'm currently leaning towards "all smash attacks deal non-integer damage", based on my experience in trying to [post=4911784]narrow down the exact damage of certain attacks[/post] (such as Lucario's u-tilt at 115%) for the Armor Project, and then later inconsistencies in damage dealt by several characters' uncharged smashes. A simple way to test a specific attack (Manondorf's f-smash, in this case) is to create a stage that has a completely-enclosed square of space, like so:
...where each "[]" represents a solid block, and the "__" represents a drop-through platform on the bottom. The whole structure should be placed above another floor somewhere, of course. Then go into training mode (so you don't have to worry about Stale Moves), place your smash test dummy in there with you, and hit them a bunch with a single attack. I've tended towards 100 times, personally, and recording percentage after every 10 attacks; just remember NOT TO PAUSE during your tests (as it would reset their damage back to 0), so you might want to set the speed to 1.5x at the beginning.
If you're masochistic, you can test this in an actual, non-Training Brawl by either a) killing Ganon between each and every hit, or b) making another box of no escape and hitting a secondary test dummy with 9 (10?) attacks between each time you f-smash your primary test dummy. But don't let me discourage you!
...though if you DO perform this test in an actual Brawl, make sure to report your results!
My general impression is that most attacks that are not smashes, are not multi-hit, and are melee attacks will deal integer damage when completely fresh.
/RtEB
