Re: generated source files, GPL and DFSG
On Sat, Jul 23, 2005 at 10:44:36AM +0100, Matthew Garrett wrote:
> Glenn Maynard <email@example.com> wrote:
> > One provided source, the other did not, and Debian considers having source
> > fundamental to having a free program.
> "Because it is, damnit"?
No, because one provided source, and the other did not, and Debian
considers having source fundamental to having a free program. If
you don't accept a simple reference to Debian's actual requirements
(DFSG#2) used to determine whether something is free or non-free
as a response to "how is one of these free and the other non-free",
then I don't know what you possibly will.
(In any case, we were trying to figure out what "source" means, not
what's more or less free.)
> If the ease of modification is equivalent in both cases, then I'd
> consider them to be equally free. If it's impractical for anyone to
> modify either, then I'd consider them non-free. "Free software" that
> provides no practical way of excercising its freedoms is not something
> that we should be supporting or holding up as an example to others.
The third sentence does not support the first two. Indeed, software
that is poorly written, and is unmaintainable as a result, is not
something that should stand as an example, and shouldn't be in Debian;
but that has no relevance to whether or not it's source.
You skipped the more relevant question: Is disassembly of a compiled
program "source" to you, if the disassembly is as usable as hand-written
assembly? I havn't seen explanations of how disassembly output, or any
forms of code obfuscation (eg. removing of comments or symbolic constants),
can possibly be seen as source code. You say "it's just as free", but
we're discussing what "source" is, not what "free" is.
You seem to be arguing that "preferred form for modification" is a poor
definition of "source" based on the fact that it doesn't permit passing
off obfuscated code (such as, perhaps, nVidia's) as "source", and that
seems to me to be one of its strengths.
(That is, from my perspective, it's self-evident that obfuscated code
is not source, and "preferred form" gets this case right. Maybe it's
self-evident to you that obfuscated code is source. If that's the case,
we may be at an impasse, having fundamentally different notions of what
"source code" means.)