I'm in the middle of a re-think on this subject. Eq is the easy one, don't use it during tracking, unless it's to use a Pultech or some other golden EQ.
Compression and/or limiting is the one I'm re-considering, as it pertains to vocal recording. Recording vocals is so much about captureing emotion, and that's such a fragile thing. Once you start telling a singer "Stand right here on this X" "Try and keep your lips from moving away from or any closer to the mic than this" or "I need you to back awy from the mic when you hit those high notes", or the worst..."That was awesome, but I need you to do it again, cause we pegged out on that take", then that emotion can fly right out the window. When I'm recording myself, I stand there in front of the mic, thinking more about the meters than about what I'm singing, and I move away from the mic when I need to, or sing off axis to it, or whatever the mic and pre need me to do to keep those meters happy. That's not working the mic....that's the mic working you! Working the mic is moving in for proximity effect and breathyness, moving out for a more open sound, etc.
For these reasons, I'm starting to warm up to the idea of using dynamic processing while recording vocals.
For other instruments, I've always said record it flat, but have a limiter in line that is there only as an insurance policy. If it never sees the threshold, perfect.
My 2 cents, RD