Subject: Re: AW: Restrictions in license
From: Ian Lance Taylor <>
Date: 26 Sep 2005 09:06:50 -0700

"Joerg Friedrich" <> writes:

> Pondering on the arguments that have come thus far, I guess I can follow
> Ians statement that OSD #6 might not make a strong case against a user
> limitation, since one could argue that the GPL also discriminates against
> commercial software product manufacturers and makes GPL'd software useless
> to them. As far as I understand, as soon as GPL'd software is introduced
> into a commercial larger works, a "commercial" software product is
> effectively turned into a "non-commercial" software product. Going one step
> beyond this, isn't this very similar to stating: "if you want to use this
> software in your commercial product, go to the owner and ask for a
> commercial license"? MySQL for example goes that way. 

I would say that you have to be more careful with your word choice.
I've worked for many years at companies which sell products covered by
the GPL.  I don't know how those could be described except as
"commercial."  Even the rationale for OSD #6 refers to "commercial
users."  What the GPL prohibits is not commercial use, but
distribution of a proprietary fork.  You could try to argue that the
GPL discriminates against proprietary software product manufacturers.
But I think it would be quite a stretch to describe that as "a
specific field of endeavor."  If we grant that, then the OSD
essentially reduces to the BSD license.

> Now Ian argues that OSD #3 states how code can be modified, and that there
> must not be any restrictions. I completely agree, but the problem here would
> not be the code but the license. Since the assumption at the start of this
> thread was that the license would be an approved OSI license, with the
> single addition of a restriction for example on the number of concurrent
> users, people could modify the code, if there was any coded restriction in
> the first place, and they could publicly redestribute it for example to take
> out all the bugs, or improve and extend the software. However, the license
> with its concurrent user count restriction would stick just as the GPL
> license sticks to the original and all derived work.
> I can understand that one could feel uneasy, since this would go against a
> "software at no cost to everybody". But how would the OSD rules prevent this
> kind of scenario, and why should they in your opinion?

I see what you are getting at.  This is a topic which has been
discussed in the past: whether the OSD requires a license to permit
people to use the software which they receive.  Most people read that
as implied--once you legally own a piece of software, it's not clear
that any license can prohibit you from using it yourself as you see
fit (i.e., any use that does not involve distribution or public
performance is unregulated).  Note that most proprietary software
licenses are careful to say that you do not own the software, but are
merely receiving a license to use it.  But OSD #1 requires that the
license permit actually selling or giving away the software.

I agree that it is a slightly grey area in theory, though not, I hope,
in practice.  Perhaps others have some better comments here.