[nSLUG] linux in the (Windows) computer lab

George N. White III gnwiii at gmail.com
Mon Sep 27 20:39:21 ADT 2010


I just returned from a two-week workshop on ocean remote sensing, and
though some of you might be interested in the approach we used to
provide a linux environment in a windows "lab" setting.   The workshop
had lectures on theory in the AM and "practicals", which I ran, in the
afternoons.   I spent a couple days setting up.   The students had
laptops (all but one running Windows) and access to a wireless network
with a Windows server and limited internet.  The course uses the NASA
SeaDAS VMware Virtual Appliance based on jeos (Just Enough OS) 8.04,
but we also used the R stats system.  Since not all the R packages
needed for the course are available in binary form, we also needed
development tools and packages.   The first thing I did was put the R
repository on the server (Windows) and make sure I could install using
a VMware shared folder.  This worked well so I went ahead and copied
an apt-mirror tree to the server along with the remote sensing data.
Moving data took about 40 hours using an external USB drive connected
to a laptop.

The first few practicals went like this:

day 1: install VMware Player and the NASA Virtual Appliance.   The
instructions on the NASA web site fall in the category of "some
assembly required".  By step 8 I you are asked to install optional
modules by extracting .tar.gz files, so after step 7 I had the
students load an image in the application and do a few basic
operations (generate a histogram, add coastlines and lat/lon grids,
etc) so they could see results a quickly as possible.

day 2: was devoted to the linux command line with the aim of at least
giving the students enough so they would feel comfortable with step 8
of the NASA instructions.  I had the students working in pairs, with
one laptop loading the GNU Linux Command Line document in Adobe Reader
and the other running the VA with a terminal in the Kate editor.  I
encouraged the students to use Kate to make logs of their work each
day, and posted my own logs to the server each day.   We created a
simple directory tree with some files, made a .tar.gz file, and then
extracted the contents before extracting the files from NASA's
packages.

day 3: we made a backup of /etc/apt/source.list and installed a
version tweaked to use the mirrors I had placed on the local server.
We used this to install a couple R meta-packages, which gave us a
complete R systems including the dev tools needed to install source
packages.

All the work was done using documentation on the NASA and CRAN web
sites with some supplementary material to help with the parts that
were unfamiliar to the students.  I hope this will be enough so they
will be able to find and install the tools when they need them in the
future.   Our lad has run dozens of courses cevering the same basic
material and the biggest issue for past attendees has been "I couldn't
get the software to work when I returned home".

After that the practicals were about using the tools for remote
sensing.  The students did individual projects which meant some needed
additional packages.   There were lots of minor problems, usually
"natural" mistakes such as omitting spaces in command lines, mistaking
minus signs for underscores, as well as spelling and capitalization
problems.  Many of these were caught when the students were working in
pairs, but it took a week before the level of typos declined.  We had
to proceed slowly and make sure every student got each step to work
before moving on.

With modern hardware, Virtual Appliances provide adequate performance
for classroom exercises.  Few sites are willing to allow native linux
installs on lab PC's, and you generally find Windows servers, so VA's
are a logical route to providing a linux environment.   I expect the
students who go on to make heavy use of the linux-based tools will
decide to install linux as a native OS, but having some previous
experience will make this much easier than if they had never been
exposed to linux.

-- 
George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia



More information about the nSLUG mailing list