Feedback appreciated :-)
Whisker[1] assists assembled applications[1] maintain correct legal
documentation[2]. A first draft[4] is command line only. The design uses
velocity to build reports from an xml meta-data document describing the
license qualities of the component assembled.
Key functions that need integration are:
* verification[5] and
* generation[6]
To integrate generation, it needs to run before final assembly.
Verification needs to run after assembly is complete. So, separating
these functions makes sense to me.
Since Maven seems to be picky about reporting, probably makes sense to
factor out the velocity reporting from the core analytics. The command
line bindings drag in extra dependencies so makes sense to factor them out.
Whisker contains yet-another-scanning framework. Perhaps one day we'll
factor out a commons scanning library...
Robert
[1] http://svn.apache.org/repos/asf/incubator/rat/whisker/trunk/
[2] For example, Apache James produces a large number of components for
the mail space. The server simply composes these components.
[3] For example LICENSE and NOTICE documents
[4] http://svn.apache.org/repos/asf/incubator/rat/whisker/trunk/legacy/
[5] checking that all the stuff in the built application has it's
license described in the xml
[6] generating LICENSE, NOTICE etc from the xml