Linux-based HP Thin Client With XFce 4

[Brian J. Tarricone]

On 09/29/04 18:05, Auke Kok wrote:
> On Wed, 2004-09-29 at 11:55 -0400, purslow at sympatico.ca wrote:
> > 040929 Olivier Fourdan wrote:
> > > On Wed, 2004-09-29 at 05:10, Bob Snyder wrote:
> > >> Olivier, are HP's enhancements going to be made available to the masses?
> > > I don't think so, it is not planned.
> > 
> > isn't there the small matter of the GPL ... (raises eyebrows)?
> 
> if they have modified xfce4/xfce-mcs-plugins to include more plugins
> they it's a direct modification of GPLv2 code (which is the license of
> xfce-mcs-plugins).

from a quick look at the relevant COPYING files, xfce-mcs-manager and
libxfce4mcs are licensd under the LGPL, so if all they're doing is making
extra modules, they're ok.  if they've modified the mcs manager or the
library, they owe us patches.

however, the xfce-mcs-plugins package (and i imagine the app-specific plugins)
are GPL, so it might be a grey area as to whether you can link a proprietary
lib and a GPLed lib to an LGPLed binary.  i don't tink that's done too often,
but personally i think the mcs manager was designed and licensed with the
ability to do this in mind.

libxfcegui4 is LGPL again.  libxfce4util is partially BSD and partially
LGPL (well, 4.0.x is all-BSD).  for the hybrid version, the collective
library is considered LGPL (since it's the more restrictive of the two), but
the BSD parts can be stripped out and used however anyone wants.

> if they haven't then you would end up in the case of kernel modules,
> where interoperability might 100% strictly technically be "derivative"
> but anyone with a code spec could write a plugin without knowing
> anything about xfce4-mcs-managerlibxfce4mcs (also GPLv2 actually). This
> would be sort of unenforcable and the kernel developers wisely have not
> gone that path (yet).

not really true.  in this case, linus torvalds has publically stated that
he won't go after _some_ devs of binary drivers.  for example, because the
nvidia driver core was developed for windows, and later ported and wrapped
for linux, linus doesn't consider it a "derivative work" in that sense.  i
believe ATI's radeon driver is a similar case.  because of a legal concept
called "estoppel" (go look it up), the act of publically saying all this
means that linus is somewhat legally-bound by his statements.

however, if a binary-only driver was written specifically for the linux
kernel, they'd probably have a case.  this is a sticky topic of discussion.
there's a lkml thread that explains all this really well, but i can't
find the link offhand.  in true slashdot style, i'll add that IANAL[1],
so i can still be wrong.

please, let's not turn this into a long drawn-out debate.  it's really not
worth it, and the on-topic portion is that HP hasn't actually violated
anything unless they've modified the mcs manager, library, or plugins.
in any case, any such modifications were probably made against the 4.0.x
version, so it's debatable as to how useful the changes would be to us.

	-brian

[1] I Am Not A Lawyer, for those of you unfamiliar with the term.