Mail Archives: cygwin/1997/02/13/10:11:30
[ jqb: Do you mind if I post this to gnu.misc.discuss (has someone already
done so ?), and we can pick it up again there ? Is there anyone
without access to gnu.misc.discuss ? ]
On Wed 12 February 1997, Jim Balter
<jqb AT netcom DOT com> wrote:
>Richard Watts wrote:
[snip]
>This is conceptual confusion; it's not a matter of Cygnus claiming
>rights over code you own, it's a matter of distribution rights.
I think it is: cygnus are claiming rights for the rest of the world
by proxy. The problem doesn't lie in the proprietory vs. free software
debate per se, but more in the associated problem that linking with
GPL'd libraries means you have to be very careful not to let even
non-computer-related trade secrets anywhere near your code (a
particularly nasty barrister might even argue that the intent of the
license is that you should upload your modifications to an ftp
site. If you don't, you're depriving the copyright holder of support
profits due to increased custom, and are guilty of a criminal offence
(in the UK, anyway). I don't think this would stand up in court,
but...). This is a recognised problem, which is why the LGPL exists
at all.
SWITCH POLITICAL
However, it's still the case that putting cygwin32 under the GPL rather
than the LGPL is likely to scare some people off due to primary
effects, and make a lot of other people's lawyers jumpy about
free software.
If it comes down to a contest between MSVC and cygwin32, MSVC will
win every time - it's supported by Microsoft, so it's unlikely to
break with an OS change, and libraries are becoming more portable all
the time. There is also likely to be strong competition from AT&T's
product when it comes on the market, and Cambridge Modula-3 (which can
be made not to use cygwin.dll, and is on less restrictive terms than
either of the GPLs if you don't use cygwin) is available now.
SWITCH NOPOLITICAL
> If your
>code can't be run without using some piece of GPL software (cygwin.dll,
>for instance), then your ownership of it does you little.
True. There's a grey area here which makes me jumpy: bash shell
scripts require GPL software, for example. So do a lot of
programs which rely (intentionally or otherwise) on GNU
bugs, or on being compiled by gcc. Linux kernel modules are
a good case in point.
I suppose the question is whether a derivative work must necessarily
contain portions of the original work, or whether it is sufficient
that the derivative work should only be comprehensible when taken
together with the original.
The former is clearly the case for books (\cite{concordances, works of
criticism}), but can be used to subvert the intent of the GPL by
distributing patches, and it's unclear how the presence of automated
tools to merge the works interacts with the situation.
> You have a
>choice: sell your code to customers on the condition that they acquire
>the GPL code code by themselves, or distribute the GPL code along with
>yours and license your software to any party for free, as the GPL
>requires.
However, this could be said to be more than a little bit silly. It
indicates that if I build an executable on my machine and ship it to a
customer site, the whole program must be licensed to
everyone. However, if I ship the source to a customer site then build
the executable, I'm safe (or am I ? Is the source a derivative work ?).
Bletch.
>
>Cygnus is now giving you another option: you can distribute cygwin.dll
>along with your code without having to give your code away, if you
>pay Cygnus for the privilege.
However, all this practically says is that it's the end user who
has to type `make'. To be frank, I think this is pretty silly...
(in fact, I think 2(b) of the GPL is a load of nonsense and should
be scrapped, or at least extensively modified. As you say, the
[L]GPL also has horrible problems with multiple-authorship
programs).
[snip]
Richard.
-
For help on using this list, send a message to
"gnu-win32-request AT cygnus DOT com" with one line of text: "help".
- Raw text -