Thinking more about this issue, I've come up with an example that I thin illustrates how we have accepted similar sloppiness in the past.
I suspect we would accept and have accepted a copyright at the top of a file that said roughly: Copyright 2003 by the Evil Empire, Inc. This software can be redistributed under the terms of the GNU GPL, version 2., with the exception that it may be linked against the OpenSSL toolkit even though doing so would be a violation of the GPL. Binaries resulting from such linking may also be distributed. That's clearly a very sloppy license. If we interpret it literally, the first party (author) grats this license to second parties. But when these parties distribute, they only grant the GPL rights to those they distribute, because that's what the GPL says happens. Such parties cannot redistribute because they cannot meet both the terms of the GPL and the OpenSSL license. Such literalism does not meet the needs of our users or the free software community. Nor I think does it reduce the legal risk to Debian in any significant manner. I strongly suspect that anyone who writes such a sloppy license especially one that seems like it is trying to be free will ask us to remove their software if they believe that we are violating their license rather than initiate legal action. I also believe that seeking damages because someone had interpreted such a license as intending to allow redistribution would be dubious in court, but IANL, so the argument about whether they would try legal action is more useful than the argument about the probable success of such action. That said, if a particularly license author is know to be litigious or is known to be picky about license texts then being more literal than usual is desirable. For example, I would recommend a very thorough reading of a University of Washington license. Similarly, if a license is written in a legalistic or formal manner, then I would recommend interpreting it in that manner. --Sam
