Re: [gNewSense-users] KFV check-in

[Bake Timmons]

> Herein lies the problem. In general, once can assume that anything in
> the kernel is LIKELY to be under GPL v2 but that requires the
> presumption that kernel.org is not voilating the GPL, which they may
> be. Whatvever Linus might say, he is NOT the author of the ENTIRE
> kernel so his opinion on what license some code is licensed under is
> as valid as yours or Bill Gates's. Only the author, or more
> specifically the copyright holder, can license software. We've found
> very CLEAR cases where the copyright holder says "This is not GPL" so
> we must assume that unlicensed code (unless CLEARLY written by Linus)
> is NOT GPL until the author/copyright holder specifies that to us.

I agree that writing the authors clearly matters more for *unlicensed*
code (as opposed to the "under the GPL" cases).  I also recall the point
marco brought up last month in
http://lists.gnu.org/archive/html/gnewsense-users/2008-02/msg00169.html

with the response from the FSF on the question of unlicensed code:

>>Unless you have some reason to believe otherwise, it is fair to assume
>>that files in a project that do not have an explicit license header have
>>the same license as the project itself. So, when you encounter files in
>>the kernel Linux that don't have license headers, you should treat them
>>as if they were released under GPLv2 only.
>>
>>If you have further questions, please feel free to contact us.

I draw a line: on "trivial" cases such as Makefiles, etc., I think we
can safely make some statement.  So far, statements have been something
like "No license, so assume to be under GPLv2."  However, a safer
statement for these cases might be something like: "No license, so
assume to be licensed compatibly with the GPLv2."

> The author's word is 100% "fact", even if this word is "Not GPL". Only
> the copyright holder can license software, so if she were to say "GPL
> version 1" or "GPL version 3" that is binding. Of course, if either
> happened, we'd have a very odd situation on our hands. It is logical
> to assume that if all the terms of the license are met, it is GPL v2,
> but assuming that without clarification means we could be violating
> the author's wishes and in some places, violating copyright law. Its
> is best to ask for clarification and remove a driver than violate the
> GPL, use an authors code against their wishes and possibly violate
> law.

OK, to me this means that the "under the GPL" language is simply
*untrustworthy* and a defect in the license (there have been others,
after all).  However, that it has survived even into GPLv3 suggests to
me that I should feel confident in exploiting it.

I have no problem contacting authors, but I also would like to believe
in the wide (albeit not universal) validity of the language of the GPL.
Having to ask suggests to me that the terms of the GPL cannot be taken
at face value.  I fail to see how exploiting the "under the GPL"
language is not worth the downside and the small risk of the downside.
After all, we are talking about a hypothetical author who would insist
that they meant GPLv1, GPLv3, or GPLv1/GPLv3, but ruled out GPLv2.

What am I missing?