Which license is that?

[Dan Ports]

On Fri, Oct 21, 2011 at 04:13:56PM -0500, Ryan Schmidt wrote:
> It's tedious to determine what license a project is under. It would be nice if I could run a script and it would figure it out for me. Do we already have one? My thoughts for how this could be done, for simple cases anyway, is to locate the license file in the tarball looking for several common filenames, normalize the formatting of its contents, and compare it with known license texts to see which one it's most similar to.

fossology (http://www.fossology.org/) claims to be such a thing,
although I've never tried it and it looks pretty complicated.

One thing that thwarts the simplest analysis you might think of doing
is that the COPYING file, if one exists, will not usually give you all
the information you need. It's pretty common to just stick a verbatim
copy of some version of the GPL in COPYING, without indicating whether
they mean that specific version, any later version, or any version at
all. It's not even particularly uncommon for packages to include a copy
of the GPL in COPYING even though the software is under the LGPL or
some other license or combination of licenses. So you really ought to
look at the copyright header in the source files.

Another approach might be to make a script that digs up the license
information from the corresponding Debian or Fedora package. I've done
this manually plenty of times. Of course, that assumes both that we can
find the corresponding package (it might be named differently) and that
they're equivalent (we might have a different version).

Dan

-- 
Dan R. K. Ports              MIT CSAIL                http://drkp.net/