i don't record anything with compression or EQ. when i'm in the studio i don't even limit. i get a feel for each musician and how they play, then i set my gain accordingly. i had some rough moments initially, but now i've gotten really good at it.
when i'm doing field work with the laptop i'm less comfortable in outside environments i bring along my compressor and do some limiting.
outside of that, i do my eq / compression / limiting during the mix process.
oh yeah, when compressing individual tracks for mixing, i decide how much compression to use by looking at the peak level and comparing it to the RSM level of the individual track. i set the threshold to the average RSM level, and set the ratio so that i get the desired amount of level reduction. i set the attack/release based on the track contents, with the release time being a multiple of the tempo in milliseconds.
when multi-band compressing for mastering, i do the same thing but i never use a ratio more than 2:1, and i seem to always have the threshold for the lows and highs set lower than the low-mids and upper-mids.
when peak limiting for mastering i seem to like it best when the threshold is set to about halfway between 0 and the average RSM level. what i mean is... if the track seems to be centered around -6db, then i'll set the threshold on the limiter to -3db so that the mastered track is 3db louder... of course, this may change because i'm better at tracking and mixing than i am mastering at this point.