[Sussex] Distros

[Steve Dobson]

Chris

On Mon, Apr 04, 2005 at 07:46:55AM +0100, Chris Jones wrote:
> On Fri, 1 April, 2005 17:18, Geoffrey J. Teale said:
> > definition is looser and is in fact Open Source rather than Free
> > Software specific (Indeed the Open Source definition is derived from
> > the DFSG).
> 
> To play the Devil's Advocate card (especially since you have a fsf.org
> address ;) I would suggest this may well be because the FSF's definition
> of Free Software, though admirable, is not that useful in the real world
> where people need software more than they need moral purity ;)

I am now convinced that the FSF's definition of Free Software (as opposed
to the Open Source definition) is the *more* useful to business.  While I
agree that sometimes RMS can be a little extreme, he is, at the end of the
day, correct.

As the proof I give you the Linux kernel.  It now includes JFS, ReiserFS
and XFS - three journaling file systems that a few years ago were 
propriety.  Would IBM, Hans Reiser and SGI have been willing to donate their
code to Linux if Linux was released under the BSD license?  I would suggest 
they would not.

If Linux was BSD licensed then IBM could have added JFS to the source and 
released it just to their customers, and the license allows that.  Under the
BSD license you are permitted to do what you like.  It isn't illegal to just
change the name and sell it as a commercial product.  Under such a license
why should any company give away their own technologies?

For companies like IBM (who's business model is based on service not product)
the GPL is an attractive license.  Their produce is not being given away, just
shared.  I can share IBM's technology with my clients, and if I need to make
a change to make it better for my clients I must share that change so IBM can
pass that improvement to those of their clients that also would benefit from
my change.

The GPL forces everyone to play fair, OpenSource License do not.  We only
have to look at the SCO v.  IBM case to see this power at work:

SCO:	JFS (and others technologies) which were earlier released under AIX
	(IBM's Unix) and are, therefore, derivative works of Unix.  We now
	own the rights to Unix, we want compensation.   Give us $4billion.

IBM:	JSF is owned by us.  We wrote it.  Why should we give you money for
	something we wrote?

SCO:	Yes, you own it, we acknowledge that.  But by releasing it first in
	AIX it is a derivative of Unix and we want a license fee from use for
	every Linux user.  The GPL is an invalid, communist, and anti-American 
	license and you had no right to release your code under it.

IBM:	You distribute Linux too, and that Linux code ships with our JFS 
	source in it.  Look, you've even left our copyright notices in the
	source files!

SCO:	So?

IBM:	We only ever released the JFS code under the GPL.  If the GPL is,
	as you claim, invalid under what license did you ship IBM code?  If
	you didn't have a license to ship the code we better sue you under
	copyright law.

SCO:	Woops.  Did we say the GPL is invalid?  No we didn't mean that, really.

SCO, having started with the stance that the GPL was invalid had to stand
up, in court, in front of the judge, and argue on the validity of the GPL.  They
need the GPL to be valid.  Not only would they be breaking IBM's (and every
other contributor's contributor to Linux) copyright, but they also ship the
GNU tools (also GPLed) with their propriety Unix offering.  Without the GPL
they have no useable product.

Thanks, in no small way to the GPL, IBM has managed to destroy most of SCO's
case against them.  Now, two years into this case, with the end of discovery
being discussed, SCO is partitioning the court to be allowed to submit a third 
amended complaint, to bring new allegations against IBM, because so little is
left of those they started with.

IBM is likely to destroy what complaints SCO has left with summary judgments 
when discovery closes.  IBM has already brought those requests for summary
judgements before the court.  But the judge dismissed them without prejudice,
saying that he didn't want to consider such matters before discovery was 
complete.  "[W]ithout prejudice" means that you are allowed to bring the
matter up later.  In fact the judge even pointed out where he thought IBM's
arguments were week, giving IBM the opportunity to fix them.

If SCO does not get to submit a third amended complaint, and IBM is successful
in it summary actions then SCO could end up in court, as the plaintive, with
only IBM's counter-claims against SCO to be resolved by a jury.  This would be
a court case where the plaintive does nothing but defend and the defendant 
does nothing but prosecute!

You say that the FSF's definition of Free Software "is not useful in the
real world".  But the OpenSource's stance on issues such software patents
is this:

    The Open Source Initiative does not have a position on whether ideas
    can be owned, whether patents are good or bad, or any of the related
    controversies. We think the economic self-interest arguments for open
    source are strong enough that nobody needs to go on any moral crusades
    about it.

I disagree.  By allowing SWpats or any other form software IP allows IT
players to build vendor lockin.  RedHat, SuSE, Mandrake, Debian, ... all
have patches that they apply to the Linux kernel.  Their clients (I'll call
Debian uses "clients" in this context) are depenant on those patches.  If
the vendors did not have to share their patches then a client would be just
as dependant on their Linux distro as Window's users are to Microsoft.

The "moral purity" of the GPL forces share and share-a-like.  A Linux user
wanting to switch distros knows that everything he is now dependant upon
*will* also be available which ever distro he switches to.  When moving
no one considers if their car is going to work as well on the roads built
by a different county council.  Why shouldn't software be as flexible?

Steve