On 12/14/20 10:54 AM, Yves Bellefeuille wrote: > > The article states that CentOS will now be "upstream" of RHEL instead > of "downstream". This is strange to me. I never thought CentOS was > upstream or downstream of RHEL; I always thought it *was* RHEL -- > perhaps a little delayed, but that's not the same as being "downstream". CentOS has always been 'downstream' of RHEL. The CentOS team rebuilt the source packages with the goal of getting as close as possible to what RHEL shipped, but it has never been 100% identical. You can do the same by pulling all of the package contents from git.centos.org and build the sources in the correct order with the correct software and the correct options to rpmbuild. Building from git.centos.org is not really hard at all; what is hard is figuring out the order and figuring out the other bits you might need that aren't necessarily on git.centos.org. Building from git is documented at https://wiki.centos.org/Sources?highlight=(git.centos.org) and you can look at an example of how I rebuilt a CentOS 8 RPM to get a non-distributed subpackage rebuilt at https://forums.centos.org/viewtopic.php?f=54&t=73376&p=314200#p314200 CentOS has never *been* actual RHEL. > It's also clear that Red Hat didn't understand the importance of the > 10-year support period. If they didn't understand it, they wouldn't offer it for RHEL. They just believe that if you need that you should pay something for it. A 10-year support lifespan, even doing a straight rebuild of the packages from RHEL, has a huge cost, and someone has to pay those costs. Should Red Hat's paying customer base subsidize those costs? (if you say 'Red Hat should pay for it' that actually means you think Red Hat's paying customers should pay for it, because that's where Red Hat's money comes from). In the case of Oracle Linux, Oracle has decided that yes, their paying support customers should subsidize the cost for those who aren't paying. Someone, somewhere, must pay the costs; in a volunteer project the volunteers typically pay the labor cost themselves, and in many cases pay the cost of the compute hardware, bandwidth, and electricity required; these are not small costs, and someone, somewhere, must pay them. If the costs aren't adequately covered, the project's deliverables suffer, and users complain. It really just boils down to a cost without a tangible return on investment. It remains to be seen if the intangible ROI was as large as the vocal reaction to the transition announcement would imply.