Can a user redistribute pay-only GPL derivative? and more...

[Ralph Corderoy]

Hi Joseph,

> > 4)      Is there any difference for interpreted script programs (ala
> > PHP) where the source essentially is the program?
> 
> Yes, slightly. It is possible that the source code supplied by the
> second party is obfuscated. To distribute that work under the GPL,
> they would be required to provide a non-obfuscated source on request,
> possibly for the payment of a small fee.

Very good point.  Section 3 has the relevant text:

    "The source code for a work means the preferred form of the work for
    making modifications to it.  For an executable work, complete source
    code means all the source code for all modules it contains, plus any
    associated interface definition files, plus the scripts used to
    control compilation and installation of the executable.  However, as
    a special exception, the source code distributed need not include
    anything that is normally distributed (in either source or binary
    form) with the major components (compiler, kernel, and so on) of the
    operating system on which the executable runs, unless that component
    itself accompanies the executable."

I've always wondered why "plus the scripts used to control compilation
and installation of the executable" isn't enforced more.  AIUI,
Stallman's intent with this part of the GPL was to be able to modify the
software that comes on a device, so just having the source that matched
the embedded binaries wasn't any use without also the means to install
your binaries built from the source after you'd modified it.

Take Sir Alan Sugar's company, Amstrad.  In the UK they sold the E3, a
videophone, which ran Linux.  It has a stereo jack socket on the back
labelled "EXT" which the manual says is for factory use only.  Turns out
it's a serial port.  In email, the E3's designer at Amstrad said they
used it during development, and also to download new firmware releases
to models sitting in the factory.  The non-GPL'd bootloader in the
device understands a protocol over that serial port to allow the
firmware to be flashed.  Doesn't the GPL mean that Amstrad should
release that protocol?  Not the bootloader's source, since it is an
in-house program developed from scratch, but document the protocol
required to replace the GPL'd code.

What if Amstrad took to only allowing "signed" firmware to be run.
Doesn't that break the GPL?  The E3's designer suggested that he'd be
happy to accept images and sign them, thereby avoiding this infraction
whilst at the same time putting enough of an impediment in the way of
the edit-compile-run cycle as to make the recipient of the source not
bother.

As it happens, the point is moot since the protocol was reverse
engineered anyway and Amstrad never got around to signing firmware for
this particular product.  They did finally release a CD of source code
for GBP25.00 but it isn't the source for the binaries on the shipped
device; they said they were too busy to come up with the right source.
Either they're lying or have abysmal software configuration management,
probably the latter.

However, it's rumoured that the Amstrad-manufactured PVRs may be running
Linux.

    http://www.advogato.org/article/911.html

Perhaps they'll try this signing get-around if people ask for the source
then so I'd be interested in learned views.

Cheers,


Ralph.