On 12/17/2013 05:50 PM, Steven Crothers wrote: > Why not run a CentOS Koji, or perhaps request access/space to the > Fedora Koji (unlikely)? > > Koji is the "standard" both for Fedora and EPEL, and I once heard it's > used internally at Red Hat as well, as far as to what extent, I have > While on the surface it sounds like a good idea, the fact of the matter is that CentOS rebuilds from already built source RPMS. This is not the normal use case for Koji, where sources, patches, and specs are its input. I say that from the point of view of actually having rebuilt the whole of CentOS 5 source RPMS for IA64 (starting point was Scientific Linux CERN 5.4 IA64, the last IA64 SLC release, and I stepwise built up through 5.9. I haven't had the time to set up automatic rebuilds or to rebuild 5.10 as yet, and it's not a high priority). Koji is overkill, as even those source RPMS that need modifying and/or rebranding aren't many. The plumbing would be nice to automate, but I found that much of what I had to do to get stuff to build wasn't easy to automate. Particularly going from 5.5 to 5.6 (I seem to remember a pretty significant delay in the official CentOS 5.6 release, and I actually found out why that was the case when actually rebuilding the packages). Manually iterating mock to build each package in sequence was necessary, and most of the time spent was spent waiting for builds to finish. The single biggest thing to remember is that RHEL is not self-hosting and would fail the normal Fedora builds-itself-from-source testing. This is a point that really cannot be overemphasized. I'm glad to see things get started a bit earlier and more publicly than with CentOS 6; kudos to the team so far for this.