Optimal license for Java projects ...

[Mitchell Baker]

It's generally agreed that a BSD-type license does allow privitization 
of a fork, including useful modifications in that fork that the original 
authors might prefer to be able to use.  Project dymanics may minimize 
this, as happens in the Apache project.  But the liense does permit a 
private fork, and this may well have been the setting that concerns Gunther.

The Mozilla Public License was pretty much written to fit into the 
setting Gunther describes.  As to the interface question:  if the 
toaster makers copy parts of an MPL file into a new file, that new file 
is governed by the MPL.  That new file can call proprietary material, 
which remains proprietary, pretty much for the reasons Gunther 
describes.   The MPL's explicit statement that it doesn't apply to the 
combined work as a whole has been very helpful to commercial entities 
which wish to partipate.  (I'm not advocating that everyone +should+ 
care about the participation of commercial entities.)

The MPL has been used by the NetBeans project, so it may be familiar 
with some of your audience.

Mitchell

David Johnson wrote:

>On Thursday 13 March 2003 09:32 am, Gunther Schadow wrote:
>
>  
>
>>- The problem of the BSD license is that it allows commercial
>>   parties to take the source code away and contribute little, and
>>   take away the freedom of their customers to use improved versions
>>   of the free code
>>    
>>
>
>You've completely misunderstood the nature of the BSD license. First, 
>commercial parties cannot take source code away any more than they 
>could take water away from an ocean. It may look like they are, but if 
>you check, the free source code is still there and the ocean isn't any 
>smaller. Second, they can't take away their customer's freedom to use 
>improved versions, because the free source code is still there and the 
>ocean is still huge.
>
>It seems to me that what you don't want is something much different. You 
>don't want a commercial company being more successful than a 
>noncommercial entity. You whole horrible scenario of MicroToast forcing 
>burnt toast on their customers can easily occur anyway, regardless of 
>license. All MicroToast has to do is to implement their own YML suite. 
>
>As a reference, I point you to BSD licensed Kerberos. For a few days in 
>history, every GPL advocate in the country was talking about how 
>horrible it was that Kerberos wasn't under the GPL. "If only it weren't 
>under the BSD", they said. But then it turned out that Microsoft didn't 
>use the BSD licensed code to begin with! They implemented their version 
>from scratch! Even RMS had to back track a bit and rewrite an interview 
>he gave(1).
>
>The point is, your scenario has never occured. But the scenario of 
>MicroToast implementing their own stuff from scratch has happened 
>innumerable times. So licenses alone aren't going to save you.
>
>  
>
>>So, what is the best open source license that Hans Hacker should
>>have used?
>>    
>>
>
>It seems to me that the best solution for Hans' paranoia is the LGPL. It 
>gives most of the benefits of the BSD license in that it can be used to 
>promote a standard YML implementation, but with just enough copyleft to 
>eliminate his paranoia.
>
>  
>


--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3