[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

New script to check license usage

In an effort to get hard data to resolve the approximately 10 outstanding
bugs that we have requesting additions to common-licenses, I wrote a
script that counts license usage by doing regular expression matches on
the copyright files unpacked in a Lintian lab.  I've pushed this into the
repository as tools/license-count (it's not installed by the package).

The results (only for the licenses I bothered to write regex checks for):

Apache 2.0             1119
Artistic               2285
Artistic 2.0             30
BSD (common-licenses)  1556
CC-BY 3.0                52
CC-BY-SA 3.0             79
CDDL                    190
CeCILL                   12
CeCILL-B                  7
CeCILL-C                 20
GFDL (any)              875
GFDL (symlink)          389
GFDL 1.2                499
GFDL 1.3                 67
GPL (any)             19893
GPL (symlink)         10116
GPL 1                  1540
GPL 2                  9073
GPL 3                  2797
LGPL (any)             7183
LGPL (symlink)         2524
LGPL 2                 4679
LGPL 2.1               3189
LGPL 3                  691
LaTeX PPL 1.3c          297
MPL 1.1                 654
SIL OFL 1.0              12
SIL OFL 1.1              55

Total number of packages: 29470

Packages that list multiple licenses increment the count for each of their
recognized licenses, so for instance some of the GPL (symlink) packages
are also counted as GPL 2 or GPL 3 packages.  The (any) counts are the
number of packages that use any of the version varients of that license.

For licenses already in common-licenses, the code only looks for them by
looking for the reference to the common-licenses path.  Lintian catches
and warns about using the license without that reference, so I think the
undercount there should be very small.

The low bar for licenses included in common-licenses by license count is
the GFDL, at 875 packages using it in some version.  None of the licenses
for which we have open bugs reach that package count.  The closest is the
MPL version 1.1, at 654 packages.

Russ Allbery (rra@debian.org)               <http://www.eyrie.org/~eagle/>

Reply to: