[Live-demo] Teaching Spatial Programming - Raspberry Pi

[Alex Mandel]

Yes, it's a different cpu architecture and could require some changes to
ensure compilation. Debian maintains an ARM build but I think QGIS or
GRASS fails to build on that arch. Ubuntu does have some support as it's
necessary to port to many tablets which now run ARM (e.g. kindle, nook,
etc) but it does not build on launchpad normally only x86 does (i386 and
amd64(x86_64)).

So for things that need compiling, we'd need to make sure they compile
or have packages for the arch. For stuff like Java it requires that
there be a jre for the arch (I'm sure openjdk from debian is there) and
that it has all the pieces people expect. As far as osgeolive goes it
would be a special VM or emulator of an arm chip or an ARM based machine
that we'd do a special ARM build on. So it would be akin to offering an
amd64 build too in terms of work, once we have a proper ARM system
somewhere.

Long term yes, I think a RasberryPi could be used for teaching
geospatial. Initially I think that might be limited to gdal/ogr, geos
and python built around those, thought the work on the QGIS android port
might be applicable.

OLPC would probably happen sooner since that x86 based (could likely
happen now).

Thanks,
Alex

On 05/31/2012 03:56 PM, Cameron Shorter wrote:
> Jim,
> I'm unfamiliar with ARM, and what impact it will have on us as software
> developers. Anything we need to know about?
> 
> On 1/06/2012 8:47 AM, Jim Klassen wrote:
>> The other key spec to the Raspberry Pi is that it is ARM rather than
>> x86 based.
>>
>> On May 31, 2012, at 16:12, Cameron Shorter<cameron.shorter at gmail.com> 
>> wrote:
>>
>>> Hello all,
>>> I've been asked whether OSGeo-Live would run on Raspberry Pi, to be
>>> used as a teaching device a.
>>> Raspberry Pi is a $25 credit card size computer. It runs Debian and
>>> only has 256 Meg of RAM. http://www.raspberrypi.org
>>>
>>> I've CCed the OSGeo-Live email list, as I expect there will be many
>>> on the list with an interest, and probably a few opinions too.
>>>
>>> The challenge will be the size of RAM. Up to version 5.0, we ran
>>> OSGeo-Live with 512 Meg of RAM, but with version 5.5 we discovered
>>> that some of the Java applications required more RAM, and we
>>> recommend at least 768 Meg RAM, and preferably 1 Gig.
>>> The problem will be all the java based applications, which are RAM
>>> intensive.
>>>
>>> You could potentially run OSGeo-Live, without invoking the java based
>>> applications and there are a C based applications for most, if not
>>> all the tasks you would likely to be teaching.
>>> You could test this be running OSGeo-Live in a Virtual Machine, and
>>> limiting the size of the RAM in the Virtual Machine to 256 Meg of RAM.
>>> You can see approximate RAM usage, and core language for many of our
>>> applications in the following spreadsheet:
>>> https://docs.google.com/spreadsheet/ccc?key=0Al9zh8DjmU_RdGIzd0VLLTBpQVJuNVlHMlBWSDhKLXc#gid=13
>>>
>>>
>>> OSGeo-Live is based upon Xubuntu (which is a small version of
>>> Ubuntu), and I note from the faq that Ubuntu is not supported by
>>> Ubuntu. That might be a problem.
>>> What you may need to do is start with a Debian Squeeze distribution,
>>> then execute the OSGeo-Live build process, probably with a few
>>> tweaks, to build a custom OSGeo-Live. In this process, you might also
>>> only select applications which will run within the 256 Meg limit.
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>