Interesting article by Lawrence Rosen (who is The Man by the way) on Linux Journal about Open Source and derivative works.
The article addresses what exactly Mr. Rosen would view a derivative work as. While this is just his opinion and not backed up by any case law in the article, he is a very respected member of the technology law industry. There also definately isn't a "IANAL" at the end of his article ;-).
When should code be considered a "derivative work?" If it's used at all in a program? So would that 100,000 line C program be a derivative of the stdio lib if it only uses one printf statement? There are clearly limits, it's understanding where these limits should lie that is the difficult part.
All the issues discussed in the article could be applied to Perl modules as well. It's an important issue that isn't considered often enough, perhaps due to the non(/anti?)-commercial nature/beliefs of the perl community (there's some good flamebait ;). There may be an impulse to just say "that's for the courts to decide" but in order to effectively oppose any questionable laws on the subject (not that questionable laws are every passed *cough* DMCA *cough*) the technical community needs to have a detailed understanding of the topic.
Enjoy the article :).