[CentOS] Upstream and downstream (was Re: What are the differences between systemd and non-systemd Linux distros?)

Sat Oct 20 23:14:21 UTC 2018
Japheth Cleaver <cleaver at terabithia.org>

On 10/20/2018 6:23 AM, Matthew Miller wrote:
> On Thu, Oct 18, 2018 at 05:52:12PM -0700, Japheth Cleaver wrote:
>> The wider EL community is trapped between a rock and a hard place
>> somewhat. If you try to direct Fedora into the needs of EL users,
>> you stand a good chance of getting told to pound stand, and that EL
>> is getting in the way of bleeding-edge progress. Traditionally,
> For what it's worth (I hope something!) I think this is an outdated fear or
> assumption. Before Fedora.next, the "default user" for Fedora was assumed to
> be an indiviual desktop user, and the overall Fedora OS offering meant to be
> one-size-fits-all but modeled to that user. That wasn't working, partly
> for the reason you identify here. Nonetheless, something like 20% of Fedora
> usage is on servers, and a lot of people work with Fedora in parallel with
> a Enterprise Linux deployment. We needed to find a place for those users to
> have a voice.
> So, Fedora Server was explicitly chartered as not just for its own sake
> (although we intend to make that true as well) but also the intentional
> upstream for downstream enterprise Linux consumers. That doesn't mean that
> every change there goes into RHEL, or is RH blessed or even Red Hat aligned
> — but the needs of EL users are *definitely* taken into account.
>> wider EL-using community. Does it want direct feedback in the form
>> of tickets? Should people form SIGs? Obviously RHEL7 is not changing
>> init systems, but where should one talk about the future?
> If this is your interest, I'd really encourage you to get more involved
> in Fedora Server. We could use your input.
This does indeed remind me of the "ring" concept, with the (perhaps 
overloaded) "Core" being something that all subsequent variations on top 
of Fedora (or Fedora-as-upstream) can use with potentially more and more 
alterations in policy, build, selection, and UX the further downstream 
you get.

The problem is that it seems like very low level decisions are and have 
been made that align most closely with the needs of the "individual 
desktop user" rather than in a more neutral manner that allows for 
meaningful distinctions *outside* of minor configs. Fedora Server can 
override Fedora configs, but it still has to deal with those Fedora-wide 
changes. Knowing at least that, for now, Fedora Server is trying to 
serve in this role is definitely encouragement to get more involved 
there, but I do fear a larger paradigm shift is involved.

Some of the Fedora-pushing is most visible in the use of Packaging 
Guidelines to implement that Fedora-specific policy; the outright 
*banning* of initscripts in RPMs (rather than allowing them to continue 
as subpackages or conditionals a la tcp_wrappers) is the ur-example, but 
there are more. Fedora inherited a lot of the moral leadership of RHL, 
but if there's question whether it can safely be considered "upstream" 
for EL (to say nothing of providing guidance other RPM-based distros), 
then I wonder if a further reorganization is necessary beyond Fedora 
into Fedora+Workstation/Server/Atomic.

Maybe if we had something above both Fedora *and* EL (whether the EL is 
RHEL or a Community *Input* ENTprise OS) which worked to enforce maximal 
downstream flexibility for its packages (rawhide specs, if you were), it 
might reduce some of this tension and provide an easier entry point for 
people wanting to get more involved in EL, but not interfere with 
overall Fedora questions. (That's really two distinct proposals there, 
but I hope my meaning comes through.)