There's not really any rules as such regarding signal routing during you mix session. For general purposes though, I'd say compression before eq. Same with reverb or delay inserts, adding a compressor after will cause the reverb/delay to respond to the compression, which might be something you don't want. Or might want, in which case you'd deliberately set it up this way. Hence me saying no rules as such.
In mixing, I always think it's better not to even listen to what other people are doing and generally try out stuff, it's always reversable anyways, and through experimentation and trial and error, you can come up with some really cool results, you may have not thought of had someone said "You HAVE to do it this way"
Mastering your tracks different, your aiming for an overall polish and sheen. It's widely agreed that the best and most effective signal routing in this process is, in it's simplist form:
Multiband compression, Parametric eq and a final limiter. If you analyse each element this makes sense. A multiband compressor ( at least 3 bands ) to even out peaks in the audio, which lets you raise the overall volume, then the eq to 'sweeten' the overall sound, and the limiter sets the final level the audio will reach.
If you want to apply things like enhancers/exciters and stereo width enhancers, slot them in after the multi band compressor and use subtley. Then shape the sound with the final eq. You could in theory try these before the compression but your then compressing the results of your 'sweetening' which kinda defeats the purpose, I always think it's best to start off by compressing the raw mixdown and working from that base up.
