How does DFSG #4 interact with texts like the Unicode data or the FHS? Is it sufficient to allow distribution of diffs along with the original sources?
It seems that from a source standpoint the answer is yes. What's unclear is how this interacts with things "built" from the sources. I honestly don't know what this would mean for FHS so I'm having a hard time answering the question. However, it seems that for the Unicode data, I could easily provide character data from them as well as diffs to that data, and have some code that would build a program based on their data plus those diffs. I think that DFSG #4 may actually give you enough wiggle room for the Unicode case.
