[Nipy-devel] FIAC data

[Jarrod Millman]

Hey,

I agree that making the initial svn download as quick as possible is a
good goal.  All uploaded data should be compressed and we should use
the existing uploaded data as much as possible.

I also think that the test data belongs in the svn tree.  The svn
repository is for developers and all the developers should be running
the tests on a regular basis.  If you move the data to an URL, you
will just have to keep downloading it over and over.  As far as
testing the URL handling functionality, we should just have one test
that grabs a dataset from our existing site.  For example,
http://projects.scipy.org/neuroimaging/ni/browser/ni/trunk/lib/neuroimaging/utils/tests/data/test_fmri.img.gz?format=raw

For now, I reduced the data directory from around 170MB to 100MB by
bzipping most of the files.  Not surprisingly, on my computer it
nearly halved the download time.  I will look into whether I can't
reduce the size more later.  For now, everyone will need to svn up.
The 2 big files are:
14M     data/FDG-de.v.bz2
33M     data/test_fmri.img.bz2
which are duplicated here:
14M     data/.svn/text-base/FDG-de.v.bz2.svn-base
33M     data/.svn/text-base/test_fmri.img.bz2.svn-base
I feel like every developer should have the test_fmri.img file, which
alone accounts for 66MB of the 100MB.  The ECAT file accounts for
another 28MB.

As far as end-users are concerned, we need to work on documentation
and packaging first.  I think that once we get to that stage we will
want to bundle the test data separate from the code.

I liked Fernando's idea about multiple test levels, but I would like
to hold off on that until we get a 0.1 release.

-Jarrod