Re: gnome-screensaver copyright issues
On Sat, Jan 23, 2021 at 11:15:01AM +0100, Bastian Blank wrote:
> Are you a copyright holder of any of the code?
No, that would be Jamie Zawinski, the author of XScreenSaver.
> Also I looked at the alleged code segment. Yes, the comments are the
> same and the code shows some resemblance, but you have not checked the
> ancestry, where does it come from, and esp from whom.
It comes from XScreenSaver, thus Jamie Zawinski, probably one of the
versions released in 2005, not 2006 as I errnoeously stated in the bug
report. The initial checkin makes it quite obvious that large
swaths of XScreenSaver code were copied into gnome-screensaver.
Almost every *.[ch] file whose name does not start start with "gs-"
seems to have come from XScreenSaver.
I wanted to pinpoint the exact revision of where this was copied from,
but apparently there is no public git repo of XScreenSaver, so that
complicates matters. Also, from his response on Twitter Jon McCann
does not appear to dispute that he copied this code. He also seems to
have copied in code from more than one revision.
For example, the current revision of
gnome-screensaver/src/gs-fade.c lacks the Zawinski notice, but was
clearly derived from src/fade.c which still had it when it was
copied in in September 2005 before it was renamed/deleted. Comparing
with latest xscreensaver-5.45/utils/fade.c (which shows a great many
verbatim commonalities) tells us that a predecessor version of it was
clearly what McCann copied.
> The MIT and GPL licenses are compatible, in a way that you can always
> add GPL-2 code to MIT code and the combined result will be GPL-2. This
> is what the gnome-screensaver author did, copyied some MIT code, which
> he is allowed to, and added GPL-2 code with his own copyright. There is
> no relicensing of the existing code, which remains MIT to this date,
> just the statement that the combined code is GPL-2.
> He also does not need to hold any copyright to the original code, just
> to retain the copyright notices, which might be lacking in this case.
That would be very good news as long as they actually do get added
back in. As I understand it redistributability pretty much hinges on
/all/ these copyright notices being in the actual release in their
original form, and they currently aren't in at least one case (and
maybe more). This is the source of my concern, and why I feel that
this might be in violation Policy 4.5.
Also fixing that will be no easy feat since someone has to actually go
through essentially the entire sources of all these 4 packages and
match up copyright notices to patch them back in.
> If it needs to ship the original license statement is up to debate,
> because the GPL-2 includes a superset of the terms, it requires
> everything the MIT license requires, but in different words.
Okay, then it's just about fixing the missing copyright notices.
> No, esp if you can't show any own copyright on the work in question.
Thanks a lot, that was very helpful and illuminating. I will leave
the bug open for now so that it can be closed when the question of the
missing copyright notices has been answered/fixed.