My chain is a little different every time (one of the symptoms/benefits of being plugin based [for now]), but certain rules usually apply:
* I like width enhancement, noise reduction, M/S mixing and de-essing (if any of them is desired/warranted) early on in the chain (process order being program material dependent).
* I usually like corrective equalization before compression.
* I usually like "sweetening" equalization after compression.
* Limiting and dither are always last, and in that order (of course).
Beyond these simple rules, lots of other things can and do happen. But there are jobs that end up being simple level sets, corrective EQ and limiting, and jobs where far more than is outlined above is called into play. The quality of the mix master is usually inversely proportional to the ammount of processing employed.
As for what makes me happy about my chain and what could be better...
What I like about my system now is that I frankly love working in the box. I can't deny it. I rarely feel the desire for knobs and instant recall lets me work with ridiculous efficiency. Nuendo is, for me, a great host for mastering, but then my methods are probably a little "unconventional" at times.
What makes me not happy is knowing with certainty that analog processing is better for some things. I can't wait to get my new STC-8/M into the chain (waiting on a new Lavry DA and RME AES-32 card to come online so that I can feed it), and a tube EQ (Massive Passive, most likely) will be the next thing to drop into the desk. I don't mind doing corrective EQ in the box at all. I think that I actually might prefer that indefinitely. But I defintely would like all of my broadband compression and "sweetening" EQ to be handled by out-of-the-box analog devices. I just think those processes will always sound best in the analog realm (watch me be proven wrong after I've already coughed up the cash for them... I keep waiting for Tritone Digital to break the "analog is still better" ceiling).