Re: Aw: Bioinformatics course in debian-med - feasibility?
Dear Andreas, Michael and Steffen,
Thank you all for your very helpful replies. I am now convinced a Debian
package is feasible. I accept this will not the most useful thing for all
users.
Here are some further details. Any big problems at this stage, please let
me know. We're going to seek funding, and I don't want to propose
something impossible.
We aim to expand the ways we release 4273pi to the following:
(1) ready-to-run SD card image for Raspberry Pi (the current situation);
(2) generic Linux version (we already provide our main Pi-based materials
as a zipfile, that anyone can download, but it's rather Pi-based); and
(3) Debian package.
(1) is what we use for teaching ourselves. It's great to give students
complete superuser control. This is how they would likely operate if they
do a biomedical PhD, for example.
(2) could easily be used by anyone with any kind of Linux; for Debian if
people want to be totally up-to-date; and even for something like OS X or
Cygwin, if someone else puts in the effort. We plan to provide the
material in editable form (as well as PDF), so it can be tailored to any
extent. Our plan to open things up with a CC0 licence will take some work
on our part, but will reduce impediments.
(3) would bring the course within the usual package-management system for
absolute ease of system-wide installation. Debian seems sensible because
it would work in widely used and specialised distributions (i.e. Debian,
Ubuntu and BioLinux).
The course consists mainly of documentation - slides for lectures, and
instructions for practical classes - plus some data. In our Raspbian SD
card image we include some packages pre-installed (e.g. emboss). In a
Debian context, then, these could just be prerequisites of our package.
This all seems like it would be unproblematic. More difficult questions,
longer term, might be in distributing everything effectively and keeping
it up-to-date. We have some plans here.
I look forward to discussing specific points at Stonehaven. Any obvious
flaws in the above, please so say before then, of course.
Thank you again,
Daniel
On 20/10/2013 19:52, ""Steffen Möller"" <steffen_moeller@gmx.de> wrote:
>Dear Daniel,
>
>> We plan to convert an open access bioinformatics course to a Debian
>> package. Because debian-med is an active, helpful and relevant project,
>>I
>> think debian-med would be a sensible route to achieve this.
>
>This is quite a compliment, I'd say. Many thanks!
>
>> The course is available as part of a ready-to-run Raspbian SD card
>>image,
>> 4273pi:
>>
>> http://eggg.st-andrews.ac.uk/4273pi
>>
>> We are willing to put in work. At the moment, I am keen to check
>>technical
>> feasibility.
>
>I think it is just fine. The Raspberry is no direct derivative of Debian,
>but this you all know better than I do.
>
>> Tasks we have in mind include:
>>
>> - More clearly identifying dependencies on external packages and between
>> components within of the course.
>
>I am not sure about what you mean, but remain assured that you only need
>to
>list the packages are identified by
>dpkg -S pathtobinaryyouexecuteorlibraryyoulinkto
>
>> - Editing the course material to suit a CC0 licence.
>
>This is not a requirement for Debian. You could be very commercial and
>we would still be happy for you using and explaining the packages.
>
>> - Preparation of a .deb.
>
>Of the course? Hm. Right. But personally I would be more interested in
>seeing the course/tutorial in the first place.
>
>> - Contribution of the .deb to debian-med.
>
>So far we have only used Wiki.debian.org to describe how to use the
>packages
>and ship the documentation that comes with the upstream source tree.
>
>> Does this seem possible?
>
>I am just not sure that this truly is what you want. In my experience the
>package would be outdated once it is shipping with stable and you will
>refer to the online version anyway. Also, to profit most from comments
>sent by your readers, you want them online and leave their insights
>somewhere.
>A Debian package sets you back much into the pre-Web 2.0 world. A wiki
>or github site is much better.
>
>> I am new to the details of debian-med. I will be at the Debian Med
>>Sprint
>> Weekend, Stonehaven (Jan-Feb 2014) which will be useful.
>
>Excellent!
>
>> At the moment, feasibility is the question. Advice on this point will be
>> welcome.
>
>Technically - yes. For a community-forming view - less so, IMHO. There is
>another point. I presume most of the work to be towards making the
>students
>understand the biology first, workflows second and command line options
>third.
>The installation of tools, i.e. the Debian-specific bits, will only make
>up
>the tiniest fraction of the effort. It would be a shame to not let others
>profit more from your work that happen to be bound to e.g. openSUSE or
>another
>friendly distro. To have as many readers as possible, and to avoid Debian
>to be special in any other way than that it saves some work for those with
>a need for those particular packages since Debian has them already, I
>would
>prefer a general approach, nothing Debian-specific.
>
>Many thanks and greetings
>
>Steffen
--
Daniel Barker
http://biology.st-andrews.ac.uk/staff/db60
The University of St Andrews is a charity registered in Scotland :
No SC013532
Reply to: