De: "Rich Megginson"
Hi,
I can't argue with that.
What I can do is present my painful experience.
I want to use fluentd (a medium sized ruby application) + a half
dozen
or so plugins packaged as separate gems.
This puts me on the hook to be the maintainer in perpetuity of 90+
ruby
RPM packages (for all of the build time and run time dependencies). Not
what I had in mind.
I can certainly relate to that, I have need of another cloud native computing foundation
project, prometheus, and what's in Fedora is awful (obsolete, incomplete, bitrotten
specs, requiring a huge number of dep packaging or repackaging, packaging guideline
updates and so on).
All I can say is someone needs to go first. Red Hat has IMHO missed the boat by focusing
on paying devs to build its own cloud offering, with the usual dev bias of NIH-ism,
unicorn-thinking, rewriting the world, building dev envs, dev containers, and not thinking
about the practical aspects of making the result available in production or making users
and ops happy in general.
And now there is a need to play catch-up, to make non Red Hat originated apps available in
the software library, in the format Red Hat customers and Fedora contributors expect
(which is *not* all the variations on custom containers which have constituted 90% of the
cloud noise on Fedora lists in the past years). And that is a huge PITA because of past
under-investments that need to be compensated first.
As a Red Hat customer, I certainly do not see the point of financing efforts that end up
with me needing to pull third party code from github and rebuild container deployment
images myself. I can do all that without paying Red Hat a penny, there is no lack of
custom container variations on the marker.
As a Fedora contributor, I do not see the point of contributing packaging apps in a format
that will help Red Hat earn money if Red Hat does not contribute a minimum to packaging
apps in Fedora. That would be Debian with the constrains of a dominant sponsor and without
the dominant sponsor earning its central place. I will do so for a while in the hope some
sanity prevails, and switch partners if it does not.
Likewise, kuddos for buying ansible tower, for releasing awx, but why is it not in Fedora?
Open but not ready to use in practice without $$$$ is open core-ism, I may as well
standardise on saltstack and migrate to saltstack paying support when my use gets critical
enough.
The cloud market is entering a consolidation phase, it's the end of paying devs to do
whatever they want in the hope of earning the jackpot, whoever focuses on grim practical
tedious necessary aspects first will leave others actors in the dust.
Regards,
--
Nicolas Mailhot