Question Regarding Derived Works

Brian Behlendorf brian at
Thu Mar 3 00:09:09 UTC 2005

On Tue, 1 Mar 2005, Chuck Swiger wrote:
> Agreed, this seems to be a very good approach which the OP might want to 
> consider.  With regard to the flaws you've mentioned, one other concern would 
> be what happens if the original source is not actually compliant with a 
> standard it references due to a bug?

What's the problem for you, the licensee?  You don't have to prove 
bugwards conformance to the buggy original source.  I think you can 
operate from the assumption that the original licensor's code is 
conformant, and that so long as your changes are conformant, you can go 
private.  If the licensor claims your derivative work is non-conformant, 
the first thing you could do is check their code (presumably there's some 
objective way to measure this - a test suite that is part of the 
work itself?) and if it's also not conformant, you can take that up with 
them.  Though, discovering the source of some non-conformance is usually 
most of the work in fixing it.

On Tue, 1 Mar 2005, Joel West wrote:
> In addition to the suggestions already offered, you could have a license 
> that says "if you change the interfaces, you cannot call it TurboSoft". 
> Brian may be being modest, but I believe that's the main difference 
> between the Apache 1.1 and (New) BSD license is to make explicit that 
> derivative works can't use your trademark.

The BSD license was most recently updated (not sure which version you're 
calling "New") to remove the "advertising clause", which made it 
GPL-incompatible and others complained about.  Apache's 1.1 and 2.0 
licenses do still have an acknowlegements clause that we believe is very 
easy to live with, and doesn't seem to be one of the sources of 
incompatibility today between Apache 2.0 and the GPL.


More information about the License-discuss mailing list