Barefoot until the garage is below about 40F, then I throw on an old pair of Vans. Not because I necessarily think they're better than anything else, but they're what I have around that's flat and easy to throw on. Seem to work pretty well though. And they stop my feet from going numb when its 30-something and drafty in the garage.
I have a pair of Vibram ten-toes but I usually don't bother dragging them out for working out- they mostly get used for windsurfing in the summer, although I do a "barefoot" run in them occasionally just to reset form an make sure I'm not depending too heavily on the cushion and support in my running shoes and using what would be an otherwise dysfunctional gait (not that I've been doing many runs since I've been focused on S&S has been a daily appointment for me).
Obvious and inherent superiority of barefoot training aside, I have thought about whether the transferability of capabilities developed using GPP like S&S to tasks demands in real life where one is inevitably going to have to perform in footwear is enhanced, at least in terms of psychological comfort, by training in some sort of shoes at least occasionally.
I heard the argument made back when I was training at a Chinese martial arts school for a few years, where some of the students would wear footwear for forms and light-contact sparring (not for "real" sparring, of course), and some wouldn't. One of the arguments made for using shoes at least occasionally, is that, if you should need to defend yourself in real life, you're probably going to be wearing shoes. And if all of your practice doing anything remotely athletic, much less explicitly martial skills, is in bare feet, you're going to feel clumsy in shoes when you least need to.