Re: All GPL'ed programs have to go to non-free
>>>>> "Adrian" == Adrian Bunk <bunk@stusta.de> writes:
Adrian> On Fri, Apr 15, 2005 at 11:58:52AM -0400, Hubert Chan wrote:
>> ... In fact, I've never looked at the gcc documentation other than
>> to look up machine-specific options and optimization flags. It's
>> easy to use gcc without the documentation.
Adrian> Simple usage might work, but as soon as you reach any question
Adrian> like e.g.
Adrian> How do I pass in a additional path to the include path of gcc?
Adrian> Which optimization levels does gcc support?
These are widely known, and can be found in many makefiles, or by asking
around.
Optimization levels is also not something that is required knowledge;
you an write a perfectly usable program without tweaking the
optimization.
Adrian> Which optimization option is best for my CPU?
You wouldn't be able to find this in the documentation anyways. The
"best" optimization depends on the CPU on the program characteristics.
If you want to find this out, you could run your program through
something like acovea.
Adrian> you are pretty lost without the documentation.
To make things clear, I'm _not_ saying that the documentation is
useless. There are some things that you probably can't find out without
reading the documentation (or the source code), such as more obscure
stuff like architecture-specific flags. *But* documentation is in a
much different class than license text. For license text, it is
*legally required* to include it with the software. But software is
still usable without documentation, and hence documentation is not a
dependency.
In fact, there's still quite a bit of software that doesn't even have
documentation (of free documentation). If gcc (or MySQL, etc.) didn't
come packaged with documentation and the only documentation available
was non-free, we wouldn't be having this discussion. We'd all just be
complaining that gcc doesn't have any good free documentation. But
people would still be able to use gcc -- just not as well as they would
if they had access to good documentation. Just like many completely
undocumented programs today are still usable, but not as well as they
would be if they had some form of documentation.
This doesn't mean that these undocumented programs are unsuitable for
distribution. This just means that someone should write some (free)
documentation. Likewise, gcc doesn't need to go into non-free just
because its documentation is DFSG non-free. It just means that someone
should write DFSG free documentation. (Or that we should modify the
DFSG so that the GFDL can be considered DFSG free, but I can't think of
any sane [1] definition of free that would allow that; obviously the FSF
thinks differently from me.)
[1] "sane" meaning it has to still have some semblance to free-ness. If
you define "free" as "everything is free", then certainly the GFDL is
free, but most people would agree that such a definition is non-sane.
Yes, I agree with you that good documentation is very useful and helps
you use the software better than if you didn't have access to good
documentation. But it is _not_ a *requirement* or *dependency*, since
it is still usable without the documentation.
I recently started looking at the GNUstep libraries, and tried my hand
at writing a program using them. Documentation does exist, but it is
highly inadequate -- it's easier to just read through the header files.
So I ended up buying a couple of non-free books. Does this mean that
GNUstep is non-free because I needed to buy some non-free documentation?
No, it just means that we need to improve the quality of the free
documentation.
[...]
Adrian> My point is:
Adrian> Non-free was going to contain mostly obscure things now that
Adrian> there are free replacements for Netscape and Acroread.
Adrian> Debian's steps of moving more and more things into non-free
Adrian> forces many users to use non-free who wouldn't do otherwise.
Adrian> Is this effect really wanted?
No. Ideally, we would have documentation that was DFSG free.
But what are the alternatives to moving GFDL documentation to non-free?
It seems you are suggesting that we could move gcc et al. into non-free
as well, and I think we both agree that that would be an even worse
solution.
We could also just keep GFDL documentation in main, but that would
violate our Social Contract as it stands today (or, rather, after Sarge
is released). We could modify our social contract again (but after GR
2004-003 [2], I doubt there would be much support for that, although
anyone is free to try). We could keep postponing the Social Contract
changes [3] indefinitely, but I don't think that would get much support
either. Or we could modify the DFSG so that the GFDL would be
considered free. Again, I can't think of a sane definition of free that
would include the GFDL, but if anyone can do it, and can convince 3/4 of
the DDs of that, then they are free to propose a GR to modify the DFSG.
The DFSG already does contain some concessions, so it may be possible to
add yet another concession so that the GFDL can be included. But I
wouldn't hold out much hope of such a resolution passing. (But feel
free to be more optimistic than me.)
[2] http://www.debian.org/vote/2004/vote_003
[3] http://www.debian.org/vote/2004/vote_004
Or we can try to convince the FSF to change the GFDL so that it is DFSG
free, or convince documentation authors to not use the GFDL. Some
people are doing this, with not much success in the former, and some
success in the latter. (I managed to convince the authors of Asymptote
[4] to change the license of their documentation from GFDL to GPL.
Asymptote is not yet in Debian, since I'm not a DD yet, but if anyone
wants to sponsor the package...)
[4] http://asymptote.sourceforge.net/
You can scream "won't somebody think of the users?" all you want.
Unless you can come up with a practical solution that would satisfy the
Social Contract, nothing is going to change.
--
Hubert Chan <hubert@uhoreg.ca> - http://www.uhoreg.ca/
PGP/GnuPG key: 1024D/124B61FA
Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Key available at wwwkeys.pgp.net. Encrypted e-mail preferred.
Reply to: