X.Org Foundation - Release Call - 3rd May 2004

[Branden Robinson]

On Wed, May 12, 2004 at 04:57:37PM +0200, Egbert Eich wrote:
> Branden Robinson writes:
>  > I have a few questions:
>  > 
>  > 1) Does Alan have a policy regarding the license of the contributions he
>  >    makes to the XFree86 CVS repository?
> 
> Why don't you ask Alan yourself instead of posting a semi-rethoric 
> question here?

I thought someone here might already know the answer, and didn't want to
pester him with a question he may have answered already, perhaps
multiple times.

Also, the question was not semi-rhetorical; it was fully sincere.

>  > 2) Does Alan use the XFree86 1.1 or X-Oz licenses?
> 
> Have you seen any indication that he does?

The lack of visual indications of usage of the new license, be they
absent from the source files themselves or from the commit messages, are
a poor indicator of the application of the new XFree86 1.1 license, as
it turns out.[1][2]

I do not feel safe relying on ambiguity under the circumstances, and in
any case it's silly to do so if I can easily obtain a clear answer, which
Alan has graciously provided.

>  > 3) Does anything in the Tungsten Graphics contract forbid people from
>  >    integrating their work into other trees as well?
> 
> ?!? .....

I'm afraid I do not understand this response.

> Brandon,

It's "Branden", BTW.

> unless you have some evidence that the X.Org tree contains code that
> is under the XFree86 1.1 licenses you should not bring this up.

I beg to differ.  Unless I can reasonably assure myself that there is no
code under the XFree86 1.1 (or X-Oz) license in the X.Org tree, I do not
feel safe using it.

How do you propose I go about reassuring myself?

> When the tree was set up it was made sure there are no such
> 'contaminations'.

By what process?  I have reviewed the list archives[3][4], albeit not
comprehesively, and have found little except merges from XFree86 CVS
after the relicensing based primarily on assumptions, at least one of
which has since been shown to be false.  (Whether Mr. Dawes intended his
comments in [1] and [2] to apply retroactively, I don't know.  Past
efforts on my part to obtain clarification from him have failed.)

> To avoid any further debate over the issue of such 'implicite licensing'
> I've turned off the automatic import of the XFree86 tree into a vendor 
> branch of the X.Org tree since those statements were made.

I think that was wise.

> We really don't need to go down this avenue.

Why not?  Either X.Org is free of X-Oz and XFree86 1.1 license
contamination and can demonstrate it, or it cannot.  If the latter,
distributors may run into trouble with licensors who use the GNU GPL and
take it seriously.

We should ensure that code borrowed from XFree86 CVS is free of
problematic licensing not just out of respect for the copyrights that
their organization and personnel can reasonably assert, but out of
respect for the many people who have built upon the -- until recently --
GPL-compatible foundation of the XFree86 source distribution.

> The tree is already in a sanitized state. There is no need to start this
> debate. 

My review of commits to X.Org CVS has made it difficult for me to
establish that it truly is sanitized, at least by my admittedly
compulsively precise standards.  There appear to be many undocumented
assumptions.  One might be "of COURSE code authored by Egbert Eich is
not under the XFree86 1.1 license, you dummy, why would you think that?"
This assumption is not explicitly documented anywhere as far as I can
tell.

Thanks to the precedent of Mr. Dawes's stated approach [1][2], I don't
know if his policy applies generally to XFree86 CVS or not, or if some
other committers to XFree86 have adopted that policy but not stated thus
anyplace I can find it.

Even assuming no one else uses Mr. Dawes's approach, what *exactly* can
I presume everyone else's policy for their commits is?

Ambiguity sucks.  Let's be clear about things, and set a better example
than XFree86.

Fortunately, Nathanael Nerode, who contributes to Debian development,
has rolled up his sleeves and prepared an almost exhaustive list of
deltas between my "sanitized XFree86" repo and X.Org 6.7.0.[5]

Anything you can do to help us sort out the status of those deltas would
be very greatly appreciated.

>  > some fonts and documentation to have licenses which fail the DFSG, but
>  > that's not really germane to this particular task -- in any case, some
>  > folks don't give two hoots about the DFSG).
>  > 
>  > What I'd *really* like to be able to offer is a proper copy of the
>  > XFree86 CVS repo, filtered as above, but I haven't had time to prepare
>  > such a thing yet. 
> 
> I think we are already far beyond this stage.

My efforts on this front are done in part for my own convenience and in
part for the benefit of the community.

If you, X.Org, or freedesktop.org don't find the resource I've prepared
to be useful, that's fine; I'm not trying to compel anyone to use it.

> Something that should be done is to identify useful patches that went 
> into XFree86 and contact the contributor to ask for permission to put 
> them into the X.Org tree also.

I vigorously agree.

> Most people probably won't mind.

I share your hope.

[1] http://www.mail-archive.com/devel%40xfree86.org/msg05906.html
[2] http://www.mail-archive.com/devel%40xfree86.org/msg05939.html
[3] http://freedesktop.org/cgi-bin/mailman/listinfo/xorg
[4] http://freedesktop.org/cgi-bin/mailman/listinfo/xorg-commit
[5] http://lists.debian.org/debian-x/2004/05/msg00362.html

-- 
G. Branden Robinson            |          Fair use is irrelevant and
Free Software Developer        |          improper.
branden at deadbeast.net          |          -- Asst. U.S. Attorney Scott
http://deadbeast.net/~branden/ |          Frewing, explaining the DMCA
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: Digital signature
Url : http://freedesktop.org/pipermail/release-wranglers/attachments/20040517/f4307a16/attachment.pgp