2008/5/23 Oron Peled <[EMAIL PROTECTED]>: > Another advantage in this kind of setup, is that when people develop > on Fedora (while integrating with RedHat/Centos) they have a looking > glass into the next RedHat/Centos release. > > This means that if they started few months ago using Fedora-8, by the > time they are ready to ship their code (after beta, field tests, > etc. ~1 year?) RedHat would ship the next RHEL based on F8+F9+F10.
And you expect customers to deploy the next RHEL as soon as it's out? Based on your explanation of the Fedora vs. RHEL relations, I'd come to the opposite conclusion - Fedora is too "futuristic" for proper production environments. And again - even for 10-20 developers, having to deal with the differences in library availability, versions, and quirks of patch variations will weight things down compared to the advantage of having some more tools around. Even installing Fedora or RpmForge packages on top of RHEL (where they don't replace relevant RHEL stuff) would seem a little more attractive than introducing a full Fedora distro into the mix, IMHO. [If you needed a Debian based system, I think the same logic would > apply: People desktop would use Debian-testing (at least for selected > packages), while final integration would be done on target > host (Debian-Etch)] I see what a headache it is to have backports around on my desktop if I want to stick to "production server setup". I'd recommend against this unless you aim to use testing/stable in production for that version of the product. --Amos (A long time Debian user who converts his workplace to CentOS-based SOE because we are forced to use them on our externally-hosted production servers).
