It really depends on the singer, and style.
But I find it useful to compress before the signal is recorded for vocals. In my experience, if I don't use compression with a vocalist; then by the time I get to a level where it doesn't clip, it's just too quiet!
In my situation, you want to get as much signal recorded as possible because my machine (Yamaha MD8) records at 16 bit with an analogue mixer.
I amplify all my mics with any one of my tube preamps. I tend to run them pretty warm so I already have that nice natural tube compression before the signal even hits the compressor. I try to set the compressor so it is only active if the track is all over the place with dynamics.
As far as EQ goes, I never do anything to any track before it is recorded. Usually good microphone technique is all you need. I don't DI anything but the bass guitar and keyboards/sampler. And usually only the base needs compression added before I record it.
If any track needs more compression when I'm listening to it with the other instruments, I take care of it later. Any gating, EQing, or FXing that's needed with an individual track after it is recorded, I do at mixdown. I try to do much of it with the busses and the two insertion paths I have. But if need be I'll make a back-up and then retrack that which needs something added/tweaked. Sadly, my machine has no undo.
