[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: The problem with gnuplot

On Tue, Aug 16, 2005 at 12:48:39PM +0200, Frederic Lehobey wrote:
> Hi,
> Disclaimer: I am not a lawyer.
> On Mon, Aug 15, 2005 at 09:58:28PM +0200, Thomas Walter wrote:
> > Here I need some light.
> > 
> > Where is a copyright break when I install from source using configure
> > options of my best choice, in this case for example 'libreadline' and
> > its best friend 'libhistory'.
> Well, I do not think there is any (depending on the precise licenses
> actually) as GPL-like licenses are considered to be enforced only
> while distributing software.  Which exactly is what Debian is for.
> See: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=322827
> > By the way, from my point of view:
> > For software in this categorie (science, heavy math oriented) it is best
> > to install always from source to profit from best optimizations for
> > underlying hardware. F.e. just think about 'atlas' and 'fftw'.
> So you think a Gentoo-like way of distributing the software would work
> around the licensing issues?
I'm not sure; if so, I guess it depends on the "end user" not redistributing
the linked code, for some definition of redistributing.  Do I have to rerun ld
on each machine with gnuplot+libreadline?  What if the machines have multiple
users?  I should probably be careful to inform my users that
/usr/{local/,}bin/gnuplot may not be copied, which is a bit difficult to understand, knowing that when I use a Debian machine, I see:

  "The programs included with the Debian GNU/Linux system are free software;"

> Instead of workarounds, could not it be easier to explain to science
> software authors the benefits of _really_ free software (I mean DFSG)
> licenses?
I agree that science software needs to be free.  If I can't apt-get
source a package, and rip it appart, see what happens if I drop a
special-case conditional, read the source, and customize it, then I
won't feel like I control the software.  Science is largely about the
ability to make assertions, and if there's stuff going on behind my
back, then my ability to do so is lessened.

> Is not the free software way of producing software a mimic of
> the way of producing knowledge and science?
Maybe; "GNU/Linux; The Peer-Reviewed OS".

> I believe most of almost-free licenses for science software are the simple
> consequence of ignorance or lack of concern for the copyright matters.
> Binary distributions like Debian cannot overlook such problems.
Maybe.  But I note that the inability of Debian to distribute a
gnuplot+libreadline package is as much a "flaw" in the GPL as it is in
the custom (?) license for gnuplot.  Each license is equally
incompatible with the other.

There are potentially many free licenses, many of which you may not be
allowed to combine in the way you want.  Potentially, this means that
you should not regard the GPL as a free license (for libraries?),
because it limits what you can do.


Reply to: