Re: More GPL questions

[David Kastrup]

Stefaan A Eeckels <hoendech@ecc.lu> writes:

> But source code as such is never a derivative work of the stuff that
> might or might not be called when the compiled program is executing.

Sure.  But the question is whether the compiling and linking is done
at the choice of the end user, or whether it is done on behalf of the
distributor:

>> For example, if one distributed an "install kit" that would
>> mechanically compile and link proprietary software with GPLed
>> libraries, it is perfectly conceivable that a court would rule that
>> going against the "licensed as a whole" demand of the GPL.
>
> This is a weird example - distributing source code of a proprietary
> product in order to compile and link it with GPLed libraries smacks
> of putting the cart in front of the horse.

It smacks of license circumvention.  If the whole kit comes with its
own (unchanged) GPLed tarball for this purpose, the distributor could
try this to claim that _he_ did not distribute a linked copy, and that
the user was free to compile, link and run the software as long as he
did not redistribute it.

>> However, it would be hard to really nail down the infringing copy
>> of the GPLed library in this process as long as the distributor
>> refrains from providing his own copies or download tools for
>> creating such copies and no library stubs have been integrated into
>> creating his product.
>
> Is a URL a "download tool"?

No.

> Does the fact that there are instructions, or even another package
> ("Build tools for Schmoo") turn Schmoo into a derivative work of the
> GPLed library it needs to be useful?  Most certainly not

Straw man.  The question is not whether the instructions turn Schmoo
without the library into a derivative of the library: of course this
is not the case.  The question is whether the actual product sold is
Schmoo _with_ the library.  If the distributor can prove that one
typical use case for a customer would be to let the code rot away
without ever compiling or linking it (indeed a typical use case for
example code), then the product does not implicitly include the
library for its completion.

> - the copyright status of the source code of Schmoo doesn't depend
> on how useful its compiled version is with or without something, but
> if the Schmoo source code is _itself_ a derivative work - in other
> words, did the author use only his own work or did he include source
> code from other authors.

The source code is not a derived work as far as I can see, but the
question is whether a mechanism created and provided for the sole
purpose of acquiring a copy to be used in order to circumvent the
requirements of the GPL does not, in itself, count as redistribution
of the person providing this mechanism.

> What we write, as programmers, is source code. All the rest is
> merely mechanical transformation, and has no effect whatsoever on
> the copyright status _of the source code_.

Sure, but the question is how the responsibilities and liabilities get
distributed.  If I hand someone a device and say "press this button",
the responsible person for an ensuing explosion is not the person
pressing the button.

>> That would be my gut feeling too (unless code passages were
>> actually taken from differently licensed example code) in this case
>> (and anyway, since the library is available identically in a
>> non-free version, one can't claim the particular protection of the
>> GPLed version), but there is also a murky borderline where things
>> become less clear even when the compiling and/or linking is done at
>> the client side.
>
> The copyright status of the source code is not affected by where,
> how and by whom it might be compiled, or the fact that it is quite
> "useless" without being compiled and linked.

Again, the question is whether the source code alone can be considered
the actual product sold.  If it is example code, this would probably
be the case.  If a convenient way of creating and using a copy of the
GPLed software for the purpose of circumventing the GPL is included,
the status of this copy and its use appears more than doubtful to me.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum