Re: agent-based simulation
Ross Boylan <ross@biostat.ucsf.edu> writes:
> On Wed, 2008-10-22 at 09:44 -0700, Scott Christley wrote:
> > On Oct 20, 2008, at 11:20 AM, Ross Boylan wrote:
> >
> > > On Sat, 2008-10-18 at 21:07 -0700, Scott Christley wrote:
> > >> Hello,
> > >>
> > >> I've been recently doing some work with the debian-med group to
> > >> package some software. There is a software package called Swarm (http://www.swarm.org
[snip]
> > > I have some interest in agent based modeling, and it seems like an
> > > appropriate topic to me. But I don't have time to do any of the work
> > > right now.
> >
> > I'm happy to do all of the packaging work, so no need for you to spend
> > time on that. Plus some work has been done already for packaging an
> > older version of Swarm, so that is a starting point for me. I'm new
> > to the Debian community so maybe I'm going about this the wrong way.
> > Essentially I'm asking if Debian-science is the right place for this
> > software, and if so what is the best way to get started?
> >
> Since those who know more haven't said anything, I'll give you my
> impressions of things.
I don't have any particular authority either. A brief view of it is the
following. More details in http://www.debian.org/doc/maint-guide/
1) Is the software useful to package and not already packaged.
> From what you've said, yes it is.
2) Can debian distribute the software ie does it meet the Debian Free
Software Guideline (DFSG)?.
Upstream says the licence is GPL, so again this should be fine.
If the code need things from non-free to compile/run then it will end
up in "contrib", not debian proper. You should also check that it
doesn't contain code not compatible with the GPL - as sometimes
happens by mistake.
3) File an Intent to Package (ITP) bug.
This avoids two people working on the same package.
4) Prepare a package
http://www.debian.org/devel/ (and in particular
http://www.debian.org/doc/maint-guide/ tell you how)
http://lists.debian.org/debian-mentors/ and the debian-science list
can help here too.
There may even be a mentors IRC channel.
If you want an SVN or Git repository for your package, I think this is
also possible - hopefully someone will be along shortly to explain how
(and perhaps add some useful wording to
http://wiki.debian.org/DebianScience/ContributingToDebianScience -
hint hint).
5) Upload the package to Debian
As you are not a debian developer, you will need to find a sponsor to
upload the package for you. They should check the package meets
Debian's quality and freeness requirements. Again, debian-mentors and
debian-scence may be able to help here).
6) Add the package to appropriate science metapackages.
7) Keep maintaining the package.
This is important - a badly maintained package is perhaps worse than
no package at all.
>
> To get a package into the official debian distribution you need a
> sponsor who is a Debian Developer and is authorized to upload packages.
> They can also help you out with the process.
>
> http://www.debian.org/devel/ is a good jumping off place for info on
> packaging things.
>
> [snip advantages of getting the package into debian]
>
> There is a list, debian-mentors (not sure of exact name) where one can
> solicit help and sponsorship. My impression from previous traffic on
> this list is that they often send people to this list for scientific
> software. As I said, yours seems to fit.
And I'd agree.
>
> This list is more oriented toward users of scientific software; a list
> specifically about packaging scientific software was recently created.
> But a lot of developer type discussion seems to happen here.
>
>
> debian-science is a mailing list and a loose collaboration focused on
> scientific software; it also has some wiki pages. I don't think it is a
> formal group or a particular customized distribution, nor do I think
> there is any central authority to say "go ahead with x."
Exactly. And I'd echo Ross's encouragement to produce a package.
>
> I am mostly an onlooker and consumer in this whole process, so all of
> the above is just a best guess, and is certainly not authoritative.
>
Nor is mine.
Good luck,
Chris
Reply to: