[SciPy-dev] Starting a datasets package, again

[David Cournapeau]

Robert Kern wrote:
>
> It works fine, provided that all of the relevant packages are registered on the
> PyPI and are installed as eggs (or with egg metadata) on user's machines.
>
>> For example, you install foo which uses faithful in some examples, when 
>> is the dependency resolved ?
>
> I would recommend that the example's dependencies be listed as an "extras"
> dependency. The setup() for, say, scikits.pyem would have these arguments:
>
>   ...
>   install_requires = ['numpy'],
>   extras_require = {
>     'examples': ['scipydata.iris', 'scipydata.oldfaithful'],
>   },
>   ...
>
> Then, if you want to be able to run the examples for scikits.pyem, you would do
> this:
>
>   $ easy_install "scikits.pyem[examples]"
>
> However, just running
>
>   $ easy_install scikits.pyem
>
> won't install the data packages (this is a good thing).
Why is this a good idea ? I guess there is a reason, but I don't see it :)

The case I am worrying about is: someone not too familiar with the whole 
thing installs pyem, and wants to go through the examples because that's 
easier than reading the doc. Then, he realizes it does not work: what is 
the error message ? Should I handle this case in my code, or is there 
some kind of mechanism to handle it automatically ?

There are already so many emails on the scipy ML (and personally, maybe 
2/3 of the emails related to my packages) because of installation 
problems, I really worry about this point. I think this hurts the whole 
numpy/scipy community quite a lot (lack of one click button "make it 
work"), and I am afraid this may be a step away from this goal.
>
> If you want a declaration from me, I would say that the surrounding text and
> code in scipydata packages should always be under the BSD license. This should
> be noted using the "License :: OSI Approved :: BSD License" classifier in the
> setup script and in a *comment* in the code following the copyright notice.
> However, the copyright notice and license should be accompanied by a note that
> the data does not fall under this license or copyright and the metadata to look
> at to find the status of the data. I'm not good at legal boilerplate, but
> something like the following would be fine, I think:
>
> # The code and descriptive text is copyrighted and offered under the terms of
> # the BSD License from the authors; see below. However, the actual dataset may
> # have a different origin and intellectual property status. See the SOURCE and
> # COPYRIGHT variables for this information.
> #
> # Copyright (c) 2007 Enthought, Inc.
> #
> # Redistribution and use in source and binary forms, with or without
> # modification, are permitted provided that the following conditions are met:
> # ..., etc.
Ok, I will prepare something in this spirit, then. Including it in 
scikits is not possible ?

David