Re: Re: Announcement: Modification to the base XFree86(TM) license.

[David Dawes <dawes@xxxxxxxxxxx>]

On Sat, Jan 31, 2004 at 03:17:42PM +1100, Benjamin Herrenschmidt wrote:
>Hi David !
>
>> On looking through the fbdev drivers in the Linux kernel source, I see
>> very few cases where license notices from XFree86 driver source are
>> included.  This means that either the license for these drivers has no
>> impact on the work you are talking about (making what you have written
>> above moot), or some authors of portions of the current Linux fbdev code
>> have violated the terms of the existing licenses by not including a
>> verbatim copy of the copyright, license notice, and disclaimer text in
>> relevant source code.
>> 
>> I would also like to echo Egbert's comments about the one-way nature of
>> your concerns.
>
>I'm mostly concerned at this point about radeonfb and rivafb. The radeonfb
>in the current mainstream kernel was written by Ani Joshi who also wrote the
>first radeon driver in XFree. So there wasn't any liencing issue at this
>point.
>However, I rewrote the kernel driver almost completely using a lot of
>informations from the XFree one as ATI is maintaining it actively.
>
>My rewritten radeonfb driver _do_ contain a copy of the licence included
>in the XFree one along with the (c) assignement.

So there is no problem as far as radeonfb is concerned.  The licence
choice for a driver always has been (and still is) the authors' choice.
The same applies for other code in XFree86.  If the authors' choice is
incompatible with your preferences, then you are free to discuss that
with the authors.  Licences like the modified XFree86 licence have
*always* been acceptable to XFree86, and some code in XFree86 already
carried licences like this prior to our latest modification.  I haven't
seen great objections to that before.  Why now?

Back to the XFree86 radeon driver, the listed copyright holders for the
bulk of that code are ATI and VA.  If those copyright holders were to
change their licenses (and whether they do or not is entirely up to
them), then you would have to approach them about such changes if they
happened to be incompatible with your requirements.

>The fact that it is mostly a one way is mostly due to the fact that the
>main problem here is seeking for HW informations. Card vendors put that
>information into XFree via drivers, we rely on this for the kernel drivers.

Speaking from my own experience, a big reason why it is one way is
because many developers err on the side of caution to avoid infecting
their code with the GPL virus, and to avoid baseless accusations from
the GPL zealots that they have illegally "stolen" GPL'd code.

Perhaps a more serious part of the "one way" problem is this very
discussion.  Has anyone considered modifying the GPL to be more compatible
with other Open Source licences rather than trying to force all Open
Source licences to be a subset of the GPL?  To me it seems a signficant
flaw to the GPL if it is incompatible with Open Source developers' desire
to receive due credit for their work from those who redistribute it.
If GPL compatibility is such a big issue to GPL advocates, then perhaps,
for their own benefit, it needs to be more accomodating than it currently
is.

David
-- 
David Dawes
developer/release engineer                      The XFree86 Project
www.XFree86.org/~dawes
_______________________________________________
Forum mailing list
Forum@xxxxxxxxxxx
http://XFree86.Org/mailman/listinfo/forum