Horst H. von Brand wrote:
>> Again, you're confused as to what the GPL does. It doesn't take any
>> such choice away. It's copyright law that does.
> But copyright law applies blindly and equally. Applying the GPL is a
> deliberate and harmful choice.
Doesn't parse.
The GPL selectively removes your choices. It makes sense for a company
selling services (like Red Hat) or companies that sell their own
enhanced version of the product (mysql, ghostscript, etc.) to apply a
restrictive license to prevent others from improving the product and
offering it in competition. It doesn't make sense to prevent those
improved versions from being available in the general case.
>>> It would be unfeasible if not technically impossible to
ship a Linux
>>> based OS containing licensed copies of all the components needed to
>>> match the functionality of commercial OS versions.
Please define "functionality of commercial OS versions". In any case, MacOS
(with its BSD core) does come awfully near if you mean closed source
userland...
Anything you are willing to pay for. No arbitrary restrictions. Yes,
Mac OS X is a reasonable option, but they don't have to distribute the
source of their Nvida and ATI drivers, do they?
And I do have a Linux based OS containing fully licensed copies of
all the
components needed to match (and in many areas, widely surpass) the
functionality I require from a "commercial OS version". It's called
"Fedora".
Can you use it to play Netflix online content?
> and components
where
> the best implementation is under someone else's copyright.
So what? The best implementation of a C compiler is under the FSF
copyright, and I have a few others lying aound here who belong to other
people,
You conveniently omitted pointers to benchmarks against Intel or Sun's
compilers.
> A
Linux
> distribution can never include those things as part of the kernel even
> if the end user is willing to pay any required license fee. The only
> scenario where this would be possible is if someone would buy
> unlimited rights for unrestricted distribution along with source. But
> there is no business model to fund such a process.
Why do you insist that the kernel must use restricted stuff?
Why do you insist that people not be free to use whatever they choose?
Even stuff like zfs that is available to everyone who wants it as a
separate component but not combined with Linux? Or that the work done
on Linux drivers not be available to someone who chooses to run
OpenSolaris? What's the point of this divisiveness?
Please note
that the kernel does include patented algorithms (RCU for one) for which it
does have your "impossible" unrestricted rights for unlimited distribution
(under GPL).
You are paraphrasing badly here. I specifically said it was possible but
under conditions that don't seem likely for everything.
> I've explained that the GPL prevents me from sharing original
work
> that links to both GPL and non-GPL libraries. Please stop say it
> doesn't prohibit me from doing that.
Copyright law (and the restrictions placed on the "other parts" by their
owners) restrict creating and distributing a GPLed whole, not GPL.
Yet it is only the GPL restriction that prevents people from being able
to get the combined work.
You are saying that e.g. GPL prevents you from sharing a chimera made
out
of Windows pieces + Linux pieces.
Or OpenSolaris, or the original BSD, or any number of other less
restrictive licenses.
It is the /Microsoft/ license which
prohibits you creating such a thing in the first place,
No it isn't - there is much more code running on Microsoft libraries
than anything else - but there is an exclusion for the standard system
libraries in the GPL anyway.
and if it did, it
would prohibit sharing at least the Windows part of it freely;
And plenty of people already have the Windows parts.
that said
monstrosity can't be shared "because of GPL" is quite off the point then.
No it isn't, that is the point.
> OK, a real example from the past: the wattcp library provided a
> TCP/IP interface for DOS programs with a 'redistributable but not
> modifiable' license and an aspi library provided scsi device access, I
> think with a 'must keep attribution' type license. Both were free of
> charge and available in source. I took the gnutar program (at the
> time not understanding that it had usurped the pdtar original or the
> implications thereof), prototyped it to compile under 16-bit DOS and
> made it use some magic device names to access the archive either on
> scsi tape or remotely over the network via rsh to another machine.
> DOS provided no way to separate these processes so all the code was
> necessarily linked into one 'work as a whole'. I thought this was a
> generally useful tool and tried to give it away, but could not because
> of the GPL restrictions prohibiting combinations with other licenses.
> And please don't try to say the problem was cause by those other
> licenses - they did not prevent anyone else from getting copies, nor
> would it have been a problem if I had started with pdtar. It was
> strictly a harmful effect of the GPL restrictions that had been
> applied to the pdtar base. I won't argue that this was illegal but it
> was certainly harmful and thus immoral to do.
Great. You can't satisfy all three licenses, and it is solely the fault of
GPL?
Yes, very specifically. If pdtar had not had the restrictive GPL applied
to turn it into gnutar, there would have been no such restriction on
redistributing a program that combined those components. How can there
be any question about that? Or that the GPL harmfully prohibits all
possible and similarly useful combinations?
> I dispute it, and would go so far as to say most free software
has
> copied much of its design from commercial/proprietary work (generally
> a good thing!), and quite a lot was actually originally developed to
> be proprietary work and later had the free license applied. Far from
> being harmful, I'd say that without the proprietary works, free
> software would barely exist and would have much less chance of future
> development. There are all sorts of variations on this theme like the
> X consortium producing free software funded by proprietary vendors.
Nonsense. Much of what the Internet is all about has always been "free
software" (even its standards, the RFCs, do count in that direction). The
propietary packages used over Internet mostly came later (or never).
Believe what you want. I don't believe Linux would exist without the
SysV interface to emulate nor *bsd without the AT&T unix the early
versions extended, X without funding from proprietary vendors, NFS
without Sun, OpenOffice without StarOffice, etc. etc.. To believe
otherwise is just too farfetched. I guess sendmail, ftp, emacs and vi
are free-software originals, though.
--
Les Mikesell
lesmikesell(a)gmail.com