We put into production a brand new service based on Fedora 5 systems. And then when CentOS 5 was available, we gradually switched to that (not by upgrading, but by buying new hardware). We needed features that weren't in RHEL / CentOS 4 and so we basically had to hope that the Fedora features would be in the next Red Hat release and that we wouldn't see unacceptable problems meanwhile. I wouldn't have even /thought/ to blame Fedora or Red Hat for any problems we encountered as a result of that gamble. On my own machines I run Fedora, but that's because I'm happy to chase the six month cycle.
What you see in that Fedora thread is definitely "I want a Pony". Too many people are basically taking the list of upsides to long term support, and the list of upsides for bleeding edge and thinking "I want everything on these lists, and nothing from the downside lists, and I don't want to pay a dime or help make it happen".
There are a lot of "OS tourists" these days. They run a Linux distro for a week, then they install SkyOS. Then they get a cracked 64-bit Vista installer, and then it's on to FreeBSD, and then a different Linux distro, and then Solaris x86, and so on. They don't really "use" any of these systems any more than a tourist "lives" in the cities they visit.
I bring the "OS tourists" up because I think they're partly responsible for this unreasonable specification of the problem. For them, every OS must have the maximum possible new and exciting stuff since that's part of the fun of these endless OS installs. Yet, since they spend so little time actually running the OS compared to getting it set up, of course they want a completely fiddle free install and zero maintenance, none of the teething problems associated with a bleeding edge Linux distro like Fedora. I believe it's impossible to satisfy such people, and pointless too, since a week after you satisfy them they will install some other OS anyway.