Re: Let's stop feeding the NVidia cuckoo
On Sat, Mar 05, 2005 at 12:40:53AM +0100, Goswin von Brederlow wrote:
> Maybe I'm too unclear. They are guidelines. As such they don't define
> what source is or what forms of 'source' are acceptable but use the
> broadest term saying just 'source'. If something is still acceptable
> as source (like having source without #define's) or not (like having a
> plain gcc -S output) has to be decided case by case.
>
> Just saying obfuscating violates DFSG#2 doesn't cut it in my
> opinion. That is far to broad a generalization to be usefull at
> all. Say the upstream author has personal references to NDA protected
> materials (e.g. "/* see page 17 of foobar */") in his source and has
> to remove them before release. Why would that make the source
> unacceptable?
>
> Having somewhat obfuscated source violates the spirit of free software
> and up to some level that can be tolerated. As long as the software
> comes under a free license (and follows it) and the maintainer is
> happy working with the source in the form it is in why should anyone
> object? The world isn't black&white but has shades of grey.
>
> Is that clearer?
I think I agree with you pretty much completely; IMO it is actually
impossible to define source in a way that is both cut-and-dried and
useful.
I also think that it would be a very good thing if we were to use our
collective discretion more often -- to say, for example, "well, you could
call this source, but it's bloody horribly ugly and painful source,
and we don't want that kind of crap in Debian" a bit more often.
There are many good reasons for doing this -- maintainability, the
chances of hiding a trojan, the fact that we want to distribute something
that we can be proud of, the fact that we want to distribute something
that our users can modify to their needs...
If someone wrote a windowing system (say, a rewrite of X) in Brainfuck
and gave you the source, it would be way way worse than any obfuscated
C source you're ever likely to see. In fact you'd probably have more
chance of making useful changes to binaries built from C code than to
Brainfuck source that big.
So let's just accept that we can't have cut and dried rules about this,
please.
Cheers,
Nick
Reply to: