As also a member of the Hackintosh community, the saying over there goes "If it ain't broke, don't fix it."
I think this applies very well to using ML. Right now I am getting consistently solid performance from ML running on 1.1.3. The builds are so stable, and everything working so well that (with the help of a new upcoming app) I have begun recommending it to my colleagues as a mature solution for professional production. While 1.2.3 might seem like a necessary leap forward, I remain cautious about upgrading. If there are no obvious advantages to be had (such as faster CF card write speed, smaller memory print that leaves more buffer available to ML, more efficient use of CPU which yields more stable GD, better thermal protection, optimized battery usage, you get the idea), then I prefer to stick with the tenet mentioned at the beginning of my post. Even if the developers assured me that 1.2.3 is 100% as solid as 1.1.3, I would still question the need to update without specific advantages to my workflow.
Of course, others have different needs and might disagree with my conservative advice. However I would caution to not succumb to the "grass is always greener..." fallacy and to obtain factual data before taking the leap.
