WordNet for DEBIAN
Upstream adopted the automake stuff that was added for versions
before 2.1. As done since WordNet 2.0 the grind source was
added to the source tarball enable building of the WordNet
database from scratch. The binary parts from the WordNet upstream
source are left out because they will be built from the text
database in the grind source.
The files lexnames and frames.vrb are not required for using
either the WordNet browser or the WordNet command-line interface. They
are remnants from WordNets past, back when the lexicographer file names
were still in a state of flux. The WordNet C library actually has both
lists compiled-in, so it doesn't need to use the files to look up the
frame text or the lexicographer file's name. But external
programs (e.g. WN::QueryData/WN::Similarity or WN interfaces in other
programming languages) might rely on them as well as they need the
So these files which are necessary to work with different language
interfaces. In a way, the Perl interface is broken.
But, it might also be pointed out that the Perl interface probably works. It's
just that the tests, which are part of the installation process, don't account
for the debian version of WordNet. (If a user forced installation, the
interface would probably work.) But the Perl interface is not currently
part of Debian and thus it is hardly possible to support this in this
I don't know providing the separate library this makes any sense but
there where questions about using the WordNet database in other
programs and so I thought, that the library could be useful.
Since version 2.0-10 the package is builded also with a dynamic library.
The hint for the necessary changes came from Guillaume Rousse
<firstname.lastname@example.org> (thanks to Guillaume).
Andreas Tille <email@example.com> Tue, 07 Nov 2006 21:11:38 +0100