Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 23 Jun 2000 11:23:29 +0900
From:      "Daniel C. Sobral" <dcs@newsguy.com>
To:        James Howard <howardjp@wam.umd.edu>
Cc:        freebsd-advocacy@FreeBSD.ORG
Subject:   Re: Response to Fatal Flaw in BSD (part 2)
Message-ID:  <3952CA21.5C234F5D@newsguy.com>
References:  <200006192016.QAA23911@rac4.wam.umd.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
I wish I had been reading my mail... Anyway, just to point out in
hindsight some flaws...

James Howard wrote:
> 
> I sent an earlier version of this to the list a couple weeks
> ago.  A lot of people liked and someone (I forget who, I am
> sorry) suggested I offer it to ZDNet who ran the original.  I offered it

Me. :-) I have OSOpinion in very low regard.

> Leibovitch blames the license for allowing Microsoft to introduce proprietary
> extensions into the protocol and claims that if Kerberos had been
> licensed under the Free Software Foundation's General Public License
> (GPL) Microsoft would have been unable to embrace and extend the Kerberos
> standard. However, Leibovitch does not get it. This was the best possible
> outcome and it was forced by the liberal license.

It wasn't the best outcome. It *would have been* the best outcome if MS
had done that. And, obviously, the license didn't "force" anything.
That's the whole point of the BSD license, anyway. :-) GPL forces
things, BSDL doesn't.

> There are four possible paths this project could have taken:
> 
> * First, Microsoft could have ignored Kerberos completely and left
>   the broader community with an entirely new standard with zero support
>   from other software in the community.
> 
> * Second, the Kerberos code could have been released under the GPL.
>   If this had happened, the Microsoft would have surely refused to
>   use the code to prevent having to reveal proprietary source. Microsoft
>   would have then reimplemented the code and still modified the protocol.

>   Had Microsoft been forced to reimplement the code, it would surely
>   contain an unknown number of bugs and compatibility issues.

These lines ought to have been removed from here, since Microsoft did
reimplement the code, by Terry's account.

> * Third, the Kerberos code could have been released under a Berkeley-style
>   license. Microsoft could have then taken the code and distributed
>   a modified version and maintained some level of compatibility with
>   existing implementations and installations of Kerberos.

Instead, you could have added here that this would result in an
implementation which is widely used and open for all to see, thus being
relatively bug-free, and avoid the 65.000 problems that seems to come
with new versions of software, and ensure that the protocol is correctly
implemented, avoiding subtle protocol interaction problems (protocol
design/implementation is a very treacherous (sp?) subject).

> * Finally, the Kerberos code could have been released under a Berkeley-style
>   license and Microsoft could have reimplemented it. This is, in fact,
>   what happened.

This is silly. You could have added dozens of cases of "Kerberos could
have been released under a XYZZY-style license and Microsoft could have
reimplemented it". You should have reworked the GPL paragraph to just
point out that Microsoft would be unwilling to use that code, and add a
final paragraph saying Microsoft decided to reimplement the code, with
all the problems that result from that.

And this whole section should have been rewritten to show that MS did
not do what Leibovich said they did, and if they HAD done it, we would
have been better off. And, above all, that GPL vs BSD has nothing to do
with it.


> The BSD license is clearly superior and offers more options for compatibility
> and interoperability because it poses no risk to business and offers
> independent developers incentive for using the code as well.

The statement that BSD license is clearly superior is always dangerous.
It isn't clearly superior to the goals of FSF. You could have said BSD
license is superior for reference implementations, as it increases the
chance of the reference implementation being actually used (and, as a
consequence, the chance of the protocol itself actually being adopted).

I'm sorry I couldn't get to you in time, I just post the above as a
feedback, so you can improve future articles.

Still, I think the rebuttal was very good, and the lack of attacks on
the rebuttal itself is a clear proof of that.

BTW, the guy defending Microsoft's implementation of Kerberos was
hillarious. :-)

-- 
Daniel C. Sobral			(8-DCS)
dcs@newsguy.com
dcs@freebsd.org
capo@the.great.underground.bsdconpiracy.org

		Windows works, for sufficently small values of "works".




To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-advocacy" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3952CA21.5C234F5D>