Re: Proposal for collaborative maintenance of packages
On Wed, 2005-12-28 at 21:27 -0200, Gustavo Franco wrote:
> skaller wrote:
> > On Wed, 2005-12-28 at 13:37 -0200, Gustavo Franco wrote:
> >> it's
> >>just a matter to publish the patches (a good PQM will do that for you)
> >>and let the others branch' maintainers do the merge when necessary.
> > Publishing patches is a serious pain in the backside
> > and a huge impediment to cooperation. I would never bother
> > to do any serious work that way. Committing directly is the
> > only effective way.
> I'm not sure that you understood my point.
I wasn't addressing your point specifically. Instead, I'm raising
another one. The whole Debian system is a serious pain and an
impediment to cooperation. Ubuntu is not much better.
I should not have to be a DD to commit to
the archive -- ANYONE should be able to commit perhaps with
some minimal registration requirements.
This is how Wikipedia works and why it is successful. With minimal
fuss I have contributed some comments and a couple of changes.
Yet with Debian I spent 6 YEARS developing some software,
it took me a month to figure out how to build the Debian
packaging .. I found a DD who is willing to not only act
as maintainer but actually joined the upstream project --
but he is very busy and there is a huge delay in upstream
changes propagating into Debian. There is no way I can
even think about fixing any bugs because of this. The package
is in Unstable and it is already two versions out of date.
I use Ubuntu .. by the time I can actually try it out on my
own system it will be so out of date I won't even remember
how that version works ;(
What you have to understand is I'm the guru on my package,
but my commitment is to IT not to Debian, Ubuntu, or some other
Linux distro. In fact, my package builds from source on all
Linux distros AND OSX, Solaris, and on Windows.
Yup, I actually use Ubuntu and it is useful to have a package
manager. I view Debian as the most significant package repository,
and appreciate the high quality.
BUT .. the entry level is way too high. ANYONE should be able
to commit to an entry level repository. Then it should be
submitted by ANYONE for inspection by an automated process.
If it passes that it goes to the next level, where a DD has
to sign it to promote it to the next level: a panel should
meet regularly to do this, and any package passed over
more than three times in a row needs a published explanation
as to why the automated processes wasn't adequate -- its a
sign of a failure in the process NOT in the package.
What I'm saying is the process is far too plagued by
administrative requirements at the entry level for the
people who actually develop the *software* to contribute --
even if they'd like to. And it is vastly too much of
an "old boys network" to join the team.
Ubuntu did try to address this but it is still a joke:
they're crying for MOTU's but when you have a look at the
crap you have to go through to maintain a Universe package
you're interested in you can see why there aren't enough.
Perhaps there is a better process than I suggest. But my
point is that the process as it stands IS A SEVERE IMPEDIMENT
So please don't talk about mailing patches. That's a joke.
It works for fixing three line bugs -- it has no hope of
ever supporting systematic changes. The only way for that
to be possible is to work directly on the repository --
one needs to commit regularly, and have others review changes,
and the feedback cycle needs to be as tight as possible --
especially when people are all around the world working at
different times and have limited opportunities to get
BTW: I know of dozens of pieces of fine software that SHOULD
be in Debian. Many were developed by academics who don't have
the time or energy to go thru the pain required to get their
stuff into Debian. Ville Laurikari's regex package is an example:
it is a Posix compliant package which is generally superior
to others: he's one of the leaders in the field, it was his
PhD topic. In other cases .. people have actually developed
Debian and Ubuntu packages .. but never contributed them
because they're UPSTREAM developers with an interest in their
own software, they're willing to support Debian and probably
desire Debian's support for their package .. but the impediments
to contributing to the archive are just too great to bother.
I would beg of those with interest in Debian to try to
figure out a way to streamline the process better, so it
can accept *grass roots* contributions more easily.
I suggest DD's and others with commitments to Debian should
work more on Policy and Tools .. and possibly acquisition
of technical resources (hardware etc) .. than uploading packages.
EG: My comp is on the net sometimes and sitting here idle.
I'd be happy if Debian used it occasionally to build binaries.
Where is the web page telling me how to advise Debian autobuilder
how to access my comp??
John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net