[SciPy-user] Re[2]: ANN: Veusz 0.5 - a scientific plotting package

[Joe Harrington]

Nicely said, Bob.  I'll add/reiterate this simple test: Say that the
"other" package were available from two (or more) independent sources
with different licenses but the same API.  Internally, those packages
could be entirely different.  Your SOURCE code is not a derived work
if you could have chosen to develop under either package without it
changing your code.  It is not even a derived work if this is
hypothetically possible (e.g., there is currently only one Qt, but
there could be another with the same API someday).  Your source code
*is* a derived work if this is not true, as for example if you had
actually taken some Qt code and used it, even modified, in yours.  In
that case, the second implementation of Qt would not have had the same
code in it, and your code would have looked different because the text
you appropriated would have looked different in the other package.

The complete running/runnable object code, of course, is a derived
work, even if it is a binary in memory, produced at runtime by an
interpreter.  If you #include something in a C program, the .o and
even post-preprocessor data would also be derived works.

--jh--