Re: How Debian Packaging practices could apply to VistA maintenance and distribution
Hi Luis,
On Tue, Jan 24, 2012 at 12:37:14PM -0500, Luis Ibanez wrote:
> ...
Thanks for the explanation (which I do not want to repeat).
> Just to clarify, here we are not talking about creating a Debian
> package for VistA. Although that is certainly something we
> want to do, and that it seems that it will be the next best target
> once we have a Debian package for fis-gtm. (we probably
> should follow up on this in a separate thread). Instead, of a
> VistA Debian package, the conversation here is about:
>
>
> Could VistA build an open source community
> after the model of Debian packaging community ?
This question is asking for a prediction.
"It is difficult to make predictions, especially about the future"
(pick your favourite person who said this from [1])
So I will not make predictions. I can only try to lay out what I would
try to do if I wanted to approach the goal above. This is a biased
opinion of somebody who is addictive to Debian on one hand but on the
other hand of somebody who managed to implement a vision with low
chances for success at the starting point to something that exists after
ten years.
The idea of Debian Med was born at LSM 2001 in Bordeaux and became
manifested at OSCHA meeting in London in 2001 (when I met Bhaskar BTW).
The idea was to stop talking about creating an operating system for
medical care but just starting to create one. I had thrown away all
attempts where people started from scratch working on their own system
(in most cases I know using and deriving Debian). So I reverted this
idea and thought we have a rock solid system like Debian which will
survive for the foreseable future (that was easy to predict) which is
flexible enough that enables *me* to change it and influence its
direction of development. At times when Fedora and OpenSuSE did not
exist this was a singular feature. Debian was a DoOcracy[2] and so
I started doing something.
If you consider the user base of people in medical care inside Debian it
was close to zero. We had some biologists and I decided to take over
the momentum coming from this group slowly reaching out into the
direction of practice management and medical imaging. Compared to the
user base of say gamers, multimedia, GIS and others we were just a bunch
of people. But the goal was to create a system for medical care which
people *want* to use because it has advantages for them. A real
advantage is not chatting about things but usable code. So while I was
observing some lists where people joined to *discuss* Free medical
Software issues I worked for the implementation and I tried to form a
team which works on the same goal. So I pinged everybody inside and
outside Debian to join our list and share our effort to add *code* to
Debian. You can perfectly follow this development when looking at the
graph of the list activity on top of[3].
I learned that we need tools to handle our packages efficiently and
enable advertising for the stuff we just have and thus invented the web
sentinel showing our stuff in a (hopefully) attractive manner. I always
intended to keep the whole Debian picture in mind and developed this
stuff for all user oriented projects called Blends. I actually
developed this Blends strategy and designed Debian Med to follow this
strategy. I admit I'm a bit concerned about other Blends who are not
flying in a way like Debian Med even if they could be potentially way
more performant considering a way larger user base featuring a way
larger package number just inside Debian. It turns out that management
of a Blend team matters. Under management I consider mails like this
patiently explaining the ideas to make people aware what they are
working for. I also consider efforts like MoM as management work and
unfortunately other Blends do underestimate the effect this management
work has onto the success of a project.
Coming back to the approach of VistA to follow the model of the
community building effort I would follow the same strategy to plug into
something that exists and will survive. I can not tell how much this
community is developed, but in case you do not want to go through the
first 4-5 hard years (see [3]) but just plug into Debian Med and
organise the packaging here you might be able to have a smooth start at
least.
The question might arise whether Debian could bear just another bunch of
very specific packages? Yes medical software is very specific to a
small user base, but Debian has a lot of those specific packages - it
just needs a doer who does the packaging[2]. As long as a package does
not show severe bugs or violates other packages, is properly maintained
and updated until now there were no voices that inside Debian is too
much medical "cruft" for an universal operation system.
Thus summarising advise would be just to start (and you did so to my
great happyness) packaging straight into Debian and see how it works.
What is the risk of this approach? Honestly I can't see anything. In
the worst case the packages would be droped from Debian (I do not see
this danger actually) and you end up with proper and policy compliant
Debian packaging which you can clone to your own repository (even if the
packages might be droped from Debian (again, I see no reason for this)
the development Vcs of Debian Med team will stay open also for external
stuff. This could be a proper base to set up a (VistA) private package
repository which people could add to their sources.list. So no harm
can happen at all.
On the other side what could you win from the completely internal to
Debian approach following the Debian Med path? Well, at some point in
time Debian will release with a complete VistA system which can be
installed for instance saying `apt-get install med-his` (or whatever you
decide). *Any* person in the world who is buying / downloading a set of
DVDs (or rather Bluerays ;-)) could theoretically do this. I mean
really anybody without the need to advertise VistA all over the world
people will just detect it when inspecting their distribution. For sure
those users who really will do are a small fraction of a percent - but
anyway I can not see any harm in this for VistA in this approach.
I told you I'm biased. It's now up to you to decide about the options
you have.
> It would seem that the human+technology infrastructure,
> + governance, that Debian has established for its own
> packaging is the proper level of support and organization
> that will be needed to take care of the natural complexity
> of VistA. In particular, Debian has found the sweet balance
> between agile behavior and formal processes, that make
> possible to keep up with the rapid innovation of the upstream
> packages, while still maintaining structure and organization.
Yes. And so it is your choice is to use this straigth or try to
reimplement it. From my biased point of view the first option is
the safer choice.
> To answer that question above we would need to get the
> combine the expertise of Debian packagers with the
> experience of VistA developers and users.
>From past ten years experience I would say that uniting under the hood
of Debian Med seems a promising option.
> I hope this topic is of interest to this list, and if so,
> we probably should make an effort to bring more
> VistA developers into this conversation as well.
Fine. Point them to the subscription page of the Debian Med list. :-)
And now back to coding
Andreas.
[1] http://www.larry.denenberg.com/predictions.html
(as a physicist I'd vote for Albert Einstein, Enrico Fermi
or Niels Bohr ;-))
[2] http://www.communitywiki.org/en/DoOcracy
[3] http://debian-med.debian.net/
--
http://fam-tille.de
Reply to: