Re: Bug #189164: libdbd-mysql-perl uses GPL lib, may be used by GPL-incompatible apps
On Tuesday, May 27, 2003, at 12:22 US/Eastern, Nathanael Nerode wrote:
First, any interface which could be used by humans is a "method of
operation". This is essentially all interfaces.
That's a good question. I think the decision only covers interfaces
that humans need to use to use the program (and all its features).
However, for a library, it is quite arguable that the API is covered.
Since dynamic linking
involves the copying of small, (usually) uncopyrightable, code
together with the use of an interface, dynamic linking incurs no
obligations, and the FSF's interpretation is quite wrong.
This is quite possible, especially since if the normally copyrightable
commands used in a menu are not copyrightable because they are a method
of operation, its possible the method names, etc., needed to dynamic
link are, too.
Second, any interface which is designed for use by humans is a "method
of operation". This is more restrictive, but does not touch the
question of secret interfaces. Suppose that passing
"-EF)#I^FMCVS)*@34%)GS234@$#7sd" on the grep command line gives access
to the secret operations -- is this the means by which a "person"
I think its probable. Even if it isn't, I doubt you could claim
copyright over the above string. Same problem with the longer (random)
If my second extrapolation is the correct one, then the documentation
status *may* be relevant as evidence for whether the interface was
designed for humans or not, though what matters is whether the GPL
program creator documented it.
I think that a program which uses command line arguments, one of the
standard ways of user interaction on Unix, is very strong evidence that
it is a method of operation. It'd be hard, I think, to argue that the
command line interface is not intended for humans.
(In fact, its other interface designed for 'people to operate it' is
interface for being statically linked with programs, so use of static
linking probably isn't covered either! Although in that case the
combination is probaly a 'derived work' and may not be redistributable;
but individuals can do the linking themselves quite freely.)
The GPL never covered use, and copyright law (at least in the US)
specifically exempts copying needed to use a computer program, so I
don't think the GPL ever prohibited private, non-distributed linking.
If my third extrapolation is the correct one, then the FSF's
library linking interpretation is valid, but we've wandered into a
remarkably fuzzy realm. Certain command-line options are rarely, if
ever, used by humans; for instance, options designed to produce better
Those options are _very_ often used by humans. Part of the UNIX user
interface is the pipe operator in every shell.
However, if the 'linkage' causes program B to rely on copyrighted
elements of program A, but the 'linkage' is not actually performed
except by the end-user, then there is no infringment on the part of the
distributor. And the GPL puts no restrictions on 'use', or indeed
creation of derived works as long as they're not distributed, so
there's no cause for contributory infringment, as far as I can tell.
I don't think there is any way to be a contributory infringer when
there is no infringement. I'm not sure if eventually dynamic linking
will become a way around copyleft licenses, but I wouldn't be surprised
if --- should it ever be tested in court --- it turns out to often be.
Disclaimer: This is not legal advice!
Very strong ditto there.