Originally Posted by GazL
I'm not really sure how to take that as the result from my sarcasm detector is coming back as 'inconclusive'. It's been a common trend in Greg's public postings over the years that he finds the long-tail of kernels that remain in-use frustrating and he often encourages people to move to the more recent versions in his release announcements. Is he really suggesting that running from Linus' tree is the "smart" choice? The Release Candidates (especially the early ones) would to me seem to be far too risky and if you only track linus' official releases that means you're going 3 months at a time without security updates and other fixes (unless you identify and back-port them yourself). Or perhaps he's just saying that dumb people like me should be using the stable kernel. Unfortunately, I'm too dumb to figure out what he is saying.
I believe the key to understanding his comment is taking the target audience into account, which seems to be kernel developers and testers. He also encourages smart people to join their ranks. He obviously sees the value of stable branches, too, otherwise he wouldn't maintain some of them.
regarding "risk": it's in the nature of software development that areas of rapid development contain bugs, and in open source they get reported, analyzed and fixed quickly. areas of rapid development are most likely new drivers and other areas marked as experimental, so probably not active in your kernel config. Call it a risk, but every day you take much greater risks in your life.
Also, due to the "mainline first" approach, fixes appear first in Linus' git repository. So if you want them ultra quickly, do a fetch every night and checkout -f FETCH_HEAD
But then you should also read the LKML...