Thu, 11 Jul 2002 18:53:41 +0200
Timothy Murphy writes:
> On Wed, Jul 10, 2002 at 10:14:30PM +0200, Frank Mittelbach wrote:
> > approach as an option (i.e. to put LaTeX under GPL) that we came to the
> > conclusion that it is not the right approach for software of a type like
> > LaTeX.
> The GPL/LaTeX issue was evidently settled long ago,
> and I wouldn't like to re-open an old hornet's nest,
> but I've seen you refer several times to the difference in kind or type
> between LaTeX and GPL-ed programs.
> I don't really see this difference.
okay, will try though I have to drop out from this discussion for a number of
days (so replies might not get answered)
> If someone put out a new version of stdio.h ,
> it seems to me it would cause exactly the same kind of chaos
> as if they put out a new version of article.cls .
> I've never come across rival versions of, say, Linux kernel files --
> except in different versions of the kernel.
the danger for the "latex kernel" is indeed not that large these days though
in it does exist. a bigger recent example was a TeX distribution (i don't
recall for sure which one, so don't want to commit slander---those who are on
tex-implementors might remember) where somebody thought that the computer
modern fonts have incorrect metric files and "improved" them so that documents
produced different line and page breaks all over the place.
now that is according to the simple license by Don Knuth (you are not allowed
to change any file unless you are Don Knuth or give it a new name --- which by
the way is a simplified version of LPPL (and which by the way is an accepted
free license by RMS) not allowed.
but coming back to your example:
there are a number of big differences between a linux kernel software and a
a) stuff is much more likely to obviously break if you make modifications,
with LaTeX is often only breaks the possibility to exchange your documents
sucessfully. now while this is very important for the majority of users it is
often of less importance or not thought through by people changing stuff (see
the whole of this discussion)
b) more importantly linux is mainly a single user system, ie once it is
somewhere in place then fine. in other words even if you change stdio.h (and
get away with it in the sense that it doesn't break your system) then building
a distribution from that is not a problem any more for those who receive
it. they can live with it easily because the system does what they want it to
if on the other hand you change article.cls (without breaking it in the sense
that it doesn't run) the users receiving it have a big problem because one of
the important features of latex (the ability to exchange documents
successfully) has been taken away from them.
> Does this danger actually arise in practice?
yes. i gave an example above, but i could give many more.