Yavor Doganov yavor@gnu.org writes:
Software is useless without a computer to compile (and/or interpret) and execute it
Yes, that's true even for software that is not executed.
while a you can read a printed manual in the park without any machine at all.
The example of printing cryptographic programs in order that they could be exported from the USA has already been raised. Another example of the folly of trying to apply different rules to a bitstream depending on its interpretation in a particular situation.
Not all written bits of digitally representable information are useful for the society to be modified, so I find it hard to believe that a math book with an invariant section "Dedicated to John Doe, my first student at the Foo University" renders the work non-free.
This is easy to present a counter-example: I might like to excerpt a useful passage from such a work and include it in mine. My work is *not* dedicated to John Doe, and I have never had students at Foo University.
Having that dedication as an unmodifiable, unremovable section attached to the passage means I must *lie* if I want to include that passage in my work. That is a non-free restriction.
Other counter-examples are possible, but it would overburden this thread to enumerate them all. Hopefully, now that it's demonstrated, you no longer find it hard to believe that such a restriction is non-free.
It would be substantially easier if judgement about freeness of a work is a boolean value, e.g. "This byte is modifiable, it is free" or "This byte is not modifiable, it is non free".
Fortunately, there is no copyright jurisdiction (to my knowledge) that allows copyright on individual bytes, so this isn't an issue.
But human brains are not mechanical parsers, and issues like these deserve serious thought; not always the easiest implementation/route is the right one.
Here we agree.
The GNU FDL, although not ideal in some circumstances, is the most suitable license for such type of works.
I disagree. The FDL was specifically designed for manuals accompanying programs. I argue instead that, because of the incompatibility between the FDL and works commonly applied to programs, a better choice instead is to apply the *same* license to both the program and the manual that accompanies it.
The new versions coming out soon will be even better.
I hope you're right.