Merging network and network-bytestring: license question

[wren ng thornton]

On 10/28/10 11:46 PM, Tyson Whitehead wrote:
> On October 28, 2010 22:48:56 wren ng thornton wrote:
>> does not compromise the original requirement to get permission from the
>> University to use their name promotionally. Combining the two as "the
>> copyright holders" or "the University, its contributors, nor Johan
>> Tibell" etc, would have the same effect as the two versions of clause 3
>> stated separately. If someone wanted to use the University's name (or
>> yours) then they'd have to get their (or your) permission in writing,
>> same as if stated separately.
>
> I would be pretty careful about touching any license wording.  It reminds me
> of a pretty serious spat that went down awhile ago because some BSD wireless
> code was "re-licensed" under GPL.  The individual who did it had thought it
> was okay as the BSD requirements were a subset of the GPL ones.

Relicensing BSD to GPL is a major change. (Though perfectly legal, in as 
much as GPL code can always make use of BSD code, and the BSD code still 
exists separately of the merger (for a while at least).)

However, taking BSD3 and merging it with another BSD3 is fairly 
straightforward. Merging BSD3 with BSD2 is even simpler, though it 
amounts to the same thing since anyone using the merged package and 
wanting to promote themselves via Johan's name must still check that the 
University doesn't care (since implicitly their code is being used in 
the promotion). This is why I said that under one reading, the merging 
of the two licenses is exactly the University's license, with Johan 
counting as yet another contributor. (If Johan's code is lifted to BSD3, 
then the same would hold the other way around, since anyone wanting to 
promote themselves via the University would be implicitly making use of 
his work in their self-promotion.)

If you want to be really picky about it then you can take the Apple 
approach and just put all the licenses in there. Though, over time, that 
will become meaningless as the two packages begin to melt together and 
the official history of any particular part becomes unknowable. This is 
part of the reason why I prefer versions of the BSD3 which refer to "the 
copyright holders" as opposed to some specific entity.


> The issue is that there are three pieces of work.  The University's, yours,
> and the combined work.  Touching the license terms makes it unclear what the
> permissions are with respect to all these parts.  Specifically, this would make
> it not clear that the name Johan Tibell is not sacred to University's code.

But this only holds for so long. Once there's some major refactoring and 
the parts of network and network-bytestring are mixed beyond 
recognition, then there is only the combined work. The University's 
portion cannot be extracted and considered under the University's 
license. The combined work must have a license of its own, and while 
this license must respect the licenses of the component works, the 
particles of the component works have no special status (unless there is 
some algorithm specifically documented as being the sole product of 
someone in particular).

-- 
Live well,
~wren