[ILUG] The old CD thread again {RHEL this time}

[Conor Daly]

On Thu, Sep 01, 2005 at 03:01:31PM -0700 or so it is rumoured hereabouts, 
Rick Moen thought:
> Quoting Conor Daly (conor.daly_ilug at cod.homelinux.org):
> 
> > > 2.  Does the RPM build process mix trademark-encumbered contents
> > > from redhat-logos and anaconda-images into various other packages'
> > > installations, during compilation?  The answer to this is almost
> > > certainly "Yes".  If so, then many more of RHEL's _binary_ RPMs are
> > > trademark-encumbered (and thus proprietary) than just redhat-logos
> > > and anaconda-images.
> > 
> > But the right to copy cited above allows non-commercial copying of the
> > trademarks (there is further commercial copying and redistribution
> > terms in the same file). 
> 
> Yes, indeed.  This is a very generous grant, which does credit to the
> company, given its product strategy -- and, of course, is what gives
> RHEL-rebuild projects like CentOS, Scientific Linux, etc. room to
> maneouver without impairing RH's branding -- as they build different
> image files, etc. into their binary RPMs.  Just in case anyone thought
> I was being critical, I'm not:  It strikes me as an ingenious and
> generally very benevolent arrangement.

It doesn't appear to be in conflict with its product strategy to my mind.
It appears to be the case that businesses (who expect to have 'license'
fees to pay) finance Red Hat's operation while ordinary geeks are not
encumbered with added restrictions to GPL software (yes, I know the _GPL_
bits are unencumbered but the distribution as a whole is...).  Even a
business is entitled to run RHEL without paying so long as they don't want
support or updates.  This fits in with the free software model of being
able to buy your support from anyone (No, it's not an explicit goal but is
implicit in the goals of free software). 
 
> I've politely argued that point with, e.g., the Progeny founders, who
> were going around trying to claim (loosely paraphrasing) that it was an
> evil proprietary-software regime, etc.  I very much disagree.
> 
> > It does encumber a GPL licensed program with proprietary data.  Is
> > that reasonable under the GPL?
> 
> I'm not sure what you mean by "reasonable under":  Do you mean
> "compatible with"?  If so, the answer is yes.  Please see below for my
> reasoning.
> 
> > Given that the rpm is a packaging method rather than the software
> > itself, it may be reasonable but appears contrary to the GPL since it
> > encumbers the recipient from redistributing the binaries.
> 
> That is not true.  Please see GPLv2 section 2, section starting "If
> identifiable sections of that work".  As long as the included non-GPL
> component in one's variant of a third-party GPLed work is separable from
> the GPLed upstream code, and isn't derivative of that code, GPL terms do
> not need to comply, to avoid violating upstream's copyright.

I didn't say _is_ contrary, just _appears_.  This is an area where an
unscrupulous vendor can make their product undistributable while complying
with the letter of the GPL.  Redistributing their product requires a large
effort of separating non-GPL components.  If the non-GPL components are
important to the operation of their product, fair enough.  OTOH, if there
is a sprinkling of encumbered images for the sake of 'privatising'
packages of otherwise GPL software, that appears to be a (perfectly
compliant) abuse of GPL.  Not that this is necessarily the case.  The
third package with a "copyright Red Hat" license (anaconda-product)
contains the file 'splash.png' from anaconda-images and is explicitely
marked "Copyright Red Hat..." suggesting that other packages contain such
images / logos might be similarly marked.
 
 
> Snipping (in a forlorn attempt at brevity) most of your grep:

Brevity?  Hah!
 
> I would ideally want to examine the particulars of all of the above, 
> since each of them leaves me in some doubt as to their terms.

Reasonable wish.  I've assumed that any 'Public License' is distributable,
anything 'Apacheish', 'BSDish' are also distributable since imposing a
distribution restriction contradicts BSD and, I believe, Apache licenses
to the point where one could not call them 'BSDish'...  I can't find a
license on the OMRON website (which, incidentally, crashed firefox) after
a brief search which started with trying to locate the word 'English'
among many japanese characters.  W3C is distributable but I didn't find
anything relating to 'W3C IPR'
 
> Moreover, your grep catches only packages that contain the string
> "licence", and doesn't preclude packages containing components with 
> proprietary terms  -- and in general is only a quick-and-dirty and
> inherently uncertain substitute for a real licence audit.

Explicitely, it catches the 'License:' field of an 'rpm -qi' output.  I
(re)discovered later the --queryformat option to rpm so I could
explicitly retrieve the 'License' field which is what I used for the
identification of packages with uncertain 'License' entries.
 
> Not that I'm complaining.  I'm just pointing out that it's nothing like
> actually auditing the licence terms within each package.

Indeed.  One thing I thought was the case (with no justification
whatsoever) was that Red Hat Inc. was being very careful to include only
free software (or was that simply un-patent encumbered stuff?) in their
distributions.  I do imagine that, for their 'professional' product, Red
Hat Inc. would have attempted to ensure that they were accurate with their
'License' descriptions associated with their rpms.  They are explicit
(somewhere) that individual license terms apply to each program/package
and that one should satisfy oneself as to their terms.
 
Please feel free to cite any of my contributions to this thread if they
would be useful...

Conor
-- 
Conor Daly <conor.daly at oceanfree.net>

Domestic Sysadmin :-)
---------------------
Hobbiton.cod.ie
 07:36:07  up 18 days, 15:58,  1 user,  load average: 0.00, 0.00, 0.00