Getting new packages into Ubuntu
Stefano Rivera
stefanor at ubuntu.com
Mon Oct 10 18:05:52 UTC 2011
Hi Sebastien (2011.10.10_18:08:16_+0200)
> - there are lot of people out there who writer softwares and have no
> interest to learn enough about Ubuntu to become a MOTU, they just want
> to reach users, they should be welcome to join as well and in a way
> which is not to difficult for them
I'm still sitting on the fence about ARB in general. I think lowering
the barrier to entry for new apps is probably a good idea. It does come
with downsides:
* We need to divert manpower to packaging these apps. On the other hand,
that's volunteer time, and volunteers can work on whatever they want
to.
* Every package needs to be explicitly uploaded to every release.
I imagine that it means that we'll start every release with a
relatively empty ARB store, and have a rush to get new apps in.
Some will then spend a month or two stabilising. Ubuntu's 6 month
release cycle may be too short for this to be an efficient process.
* No obvious approaches to handling security issues or bug reports yet.
I got a single report for my ARB app, by a user who found the source
and hunted me down. (And I haven't dealt with it yet, eep)
* It doesn't deal very well with libraries that aren't in Ubuntu. (And
with the vague proposal of having a tiny Ubuntu core, main, without
universe, this becomes a much larger problem). They need to be bundled
with every app. This poses security problems, even if it does make the
app author's lives easier.
> - you were recently complaining as well about the number of packages
> that see one upload and stop being maintained that we have to fix then,
> do we want those in the main archive because they attract people or
> would they be better suited in extras?
Yes, this would help with that. I seem to remember an earlier proposal,
where, if an app was still popular after a release or two, it would be
strongly encouraged to be included in Universe / Debian.
> - locking upstream softwares to our release cycle just don't fit, it's a
> best un-natural and create extra work, it often means that users get
> outdated softwares or versions that upstreams want to replace
That is another reasonable advantage.
> One other way would be perhaps to stop freezing universe at release and
> to let softwares elvolve in a least strict way...
I'm sure there'd be people who'd appreciate that (it sounds rather
ports-ish), but I'm already concerned about the stability of Universe as
it is (MOTU is rather understaffed right now).
Yes, I also got involved in Ubuntu because I wanted to get a
(particularly minor) app, and all its dependencies in. I had also been a
Debian user for a decade or so, and had always intended to get more
involved in the development side of the distributions, so I might have
been more naturally drawn in than others. But I'm pretty sure that if
ARB had been available then, I would have used it, rather than sticking
my nose into #ubuntu-motu and asking where I could help out.
SR
--
Stefano Rivera
http://tumbleweed.org.za/
H: +27 21 465 6908 C: +27 72 419 8559 UCT: x3127
More information about the ubuntu-devel
mailing list