Subject: Re: Licenses vs. public domain
From: Brian Behlendorf <>
Date: Sun, 28 May 2000 23:22:01 -0700 (PDT)

On Sat, 27 May 2000 wrote:
> > - Successful near-public-domain projects.  The Apache license isn't all that
> >   far from simply being public domain, yet the Apache project hasn't suffered
> >   from lack of contributions.
> Apache is BSD.  It's an "attribution required" license.  Essentially
> serves to keep people aware of who wrote the source.  PD needn't do
> this.

Actually, the more important aspect of attribution required, at least to
me, is the name issue.  Anyone can fork the code base, but they can not
use the name Apache.  They can not call their product or derivative
Apache++, ApachePro, TurboApache, Apache For Beginners, etc etc.  They are
welcome to say, "Based on the Apache web server" or "includes the Apache
server", but derivatives don't get to take advantage of the name we've
built.  We also generally say no when people ask if they can start up a
company with Apache in its name.

Certainly this is rendered more valuable now that the name is
pretty well known and the core server itself is a solid product (rather
than a collection of libraries and code in need of others to do
'distributions').  There are downsides; I don't necessarily believe that
Linux would have been "better off" if Linus had forbidden the use of the
name Linux in the names of products using the kernel, and the companies
promoting them, for some definition of "better".  I do think we'd see less
confusion in the marketplace as to what Linux actually is, as well as less
divergence between distributions.

>  - Software licensing addresses only part of the concern of free
>    software, and specifically doesn't address protocol standardization
>    or other compliance certification, which should be handled through
>    other means such as standards organizations and applicable trade or
>    certification marks.

Don't necessarily agree.  I actually think the Sun Internet Standards
Source License is pretty cool - it basically says

  a) if you stick to the reference standards this code implements, you
     can create proprietary derivatives
  b) if you want to deviate from the standard, you must publish your
     changes in the form of documentation and a reference implementation.

It does make testing for conformance must messier than, say, the GPL does,
but aside from that it's a pretty interesting way to get the flexibility
benefits of BSD and yet still prevent circumstances like the recent
Microsoft/Kerberos deal.