Re: Are there problematic infrastructure or processes in Debian?
Raphael Hertzog <email@example.com> writes:
> Debian's infrastructure and processes have grown organically over the
> years, with all the strengths and weaknesses that it implies. Sometimes
> it's a good idea to step back and look whether some of those need
> to be amended/replaced/dropped/etc.
> Based on your own experience, which infrastructure(s) or process(es) would
> benefit from significant changes?
> Are there infrastructures or processes that we're (still) lacking and that
> could make a significant difference in the work of Debian's contributors?
As far as infrastructures go, what I find a bit troublesome is that our
tools are sometimes too diverse: too many languages, too few people to
understand and improve them. This is also a project-wide problem of not
being able to make use of our human resources better.
This, in turn, leads to situations where some of our tools look like
they're stuck in the past millennia, which is quite a bummer when it
comes to attracting new contributors, especially when said tool is
something they'll see early on. (Yes, I'm talking about the BTS, which
is a terrific thing, and I wouldn't trade it for anything else, but from
a usability point of view, it is behind times. It would help
tremendously if we had more people working on it, as one person can't
cover all aspects.)
To attract new people, we need a bit more than technical excellence, we
need to impress them, and impress them fast. This - as blasphemous as it
may sound - may require our user-facing tools to look nice, and be
friendly to newcomers. A big problem with parts of our infrastructure is
that this is not the case.
On the topic of processes, I'd emphasize mentoring & sponsoring. This is
where we're... well... I'll be honest: we suck at it, in general. We do
have some amazing teams that do this well, we have a lot of it going on
in the background, hardly visible. BUT, we also have a lot of problems,
and they are sadly far more visible, which can easily drive off new
In my experience, mentoring and sponsoring over the internet rarely
works effectively, due to a whole lot of things, including but not
limited to language barriers, lack of knowledge (on both sides, but in
different areas), noise, lack of time, impatience and so on. What does
work, as far as I saw is meeting and talking in person. Things can go a
lot smoother and faster in real life, as there is the personal
If we could have "mentoring sprints", that would be very useful, in my
opinion. (Encouraging & empowering local teams to help this cause
further would be an important thing.)
Another area we're lacking in is recruiting non-packaging
contributors. I believe that local events, much similar to the
aforementioned mentoring sprints would be a great way to reach out to
more people, be them technical or not. Show them how Debian makes a
difference, *impress* them, and we're halfway there.