Mailinglist Archive: uyuni-devel (1 mails)

< Previous Next >
[uyuni-devel] About Uyuni development decisionmaking
  • From: Neal Gompa <ngompa13@xxxxxxxxx>
  • Date: Wed, 30 Jan 2019 00:04:18 -0500
  • Message-id: <CAEg-Je_GRFprqRiexvmCQgfyQYVUwTuUXZiFqybPAHeWK0U=_g@mail.gmail.com>
Hey all,

So I've been following the development of Uyuni in parallel to the
Spacewalk development for the past few months. I've been glad to see
that both projects are showing a steady state of development, and
Uyuni development is evenly paced.

However, there have been some (IMO) major disappointments in how the
development process for Uyuni is going, which all seem to stem from a
foundational issue: It is fundamentally impossible for me to figure
out what is going on as you guys are developing and merging in code.

The main "symptoms" of this opaque development process:

* Pull requests are referencing issues that exist in a private
repository (SUSE/spacewalk), which presumably is the SUSE Manager
codebase. This means that many pull requests have no context for
external (i.e. non-SUSE) contributors to review.

* Changes are being made with no obvious indicator of why they are
happening, and the way the developers are committing to the codebase
makes it impossible to divine any useful understanding. The style of
just slamming out change after change with inscrutable commit messages
really makes it hard to figure out what the meaningful change even is!

* To my surprise, there is apparently an RFC process for figuring out
major decisions. Unfortunately, external contributors have no
opportunity to weigh in and help with those decisions. This is because
that process is completely private and not open for public discussion.
This means that by the time external contributors can react, it's
already been implemented and there's no time anymore, which means
we're stuck with a potentially bad decision due to incomplete
information.

Each of these things have happened during my time observing the Uyuni
project. That said, it's not like Spacewalk is too much better, but we
should aim for better no matter what!

This most recently was a problem for me because I had been slowly
examining the effort required to run Uyuni on Fedora and CentOS and I
kept getting broken by changes done before I had any time to react.
The most recent one completely broke everything for me[1], and had I
known they were trying to decide on an approach to solve that, I would
have given them a better strategy that would have worked on
CentOS/RHEL 7, Fedora, openSUSE Leap 15.0, and SUSE Linux Enterprise
15. However, I didn't know until it was literally about to be merged,
which is a terrible way to find these things out. This change exhibits
the characteristics of all the problems I described earlier, because
the way changes were done means that reverting this and doing it right
will be extraordinarily painful _for me_, and it shows some
callousness for the community, because we have no opportunity to be
part of the project.

Julio and Pablo (cc'd to this email) have already affirmed to me that
they'll reconsider their approach so that this is unbroken. But this
development process needs to change. It's too daunting for me to track
and work with (and I'm pretty experienced at this stuff!). I worry
that it's even scarier for people less versed in these kinds of
projects.

I know that in a lot of projects, I wind up being the token "community
guy". And I'm okay playing that role for a while, if I can help make a
project become good enough that it can start attracting a community to
sustain itself. I've done this for a number of projects over the
years, and I'm okay with doing this. But I don't want to be only
community person forever, and the only way that will improve is if the
development process improves.

All the best,
Neal

[1]: https://github.com/uyuni-project/uyuni/pull/527



--
真実はいつも一つ!/ Always, there's only one truth!
--
To unsubscribe, e-mail: uyuni-devel+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: uyuni-devel+owner@xxxxxxxxxxxx

< Previous Next >
List Navigation
This Thread
  • No further messages