This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [commit/6.2] Fix lib (C)s; Was: src/gdb/testsuite ChangeLog lib/insight-suppor ...
- From: Daniel Berlin <dberlin at dberlin dot org>
- To: Andrew Cagney <cagney at gnu dot org>
- Cc: gdb-patches at sources dot redhat dot com,Michael Elizabeth Chastain <mec dot gnu at mindspring dot com>
- Date: Mon, 19 Jul 2004 21:41:29 -0400
- Subject: Re: [commit/6.2] Fix lib (C)s; Was: src/gdb/testsuite ChangeLog lib/insight-suppor ...
- References: <20040719223324.0E8684B104@berman.michael-chastain.com> <40FC704C.6070205@gnu.org>
On Jul 19, 2004, at 9:07 PM, Andrew Cagney wrote:
Given that the boilerplate "work" is stolen from COPYING and that
has a
(C) of 1989,1991 why should you not instead be adding those dates
(and have those dates through out all of GDB's files)?
If I had infinite time, I would.
Lets consult the resident expert then :-)
http://sources.redhat.com/ml/gdb-patches/2004-07/msg00211.html
Daniel,
the patch changes the (C) from Red Hat to FSF. The year 2003 was
added as a change was made then. Michael's asking that 2004 be also
added as that's the year.
If the only thing that changed was the copyright notice or license
text, i wouldn't do it, because you haven't changed any of the actual
copyrighted work, and thus, have not created a new derivative work.
Though some sufficiently anal retentive lawyer might tell you
otherwise.
It all comes down to whether you consider the license part of the
original work or not (since that is what you changed, if it's part of
the work, than you've made a derivative).
Let me explain why :
Copyright notice dates are dates of first publication, not dates of
change.
In our (distributed free software development) case, they are somewhat
similar.
This is because
1. You are supposed to use the date of first publication of each
derivative work in that derivative works copyright notice (In other
words every new derivative work gets a new publication date added to
the copyright notice.)
and
2. Every time we make a change to actual code, and publish it to the
public cvs server (and the web, and the ftp server), and thus, the
world, we are effectively creating a new published derivative work as
of the date/time you commit the change.
Thus,
3. the copyright date gets updated, because you've created a new
derivative work, and are publishing it, and thus, add the new date of
publication to the copyright notice.
So if you consider the license part of the protected work, and thus,
changing just the license text to be creating a new derivative work,
then you need to update the copyright date.
If you don't, you don't need to update the copyright date.
Given all of that:
Unless you think that changing only the license text creates a new
derivative work (i don't, because i doubt that legalese is a
protectible part of the work), you don't need to update the copyright
date.
As i said, some sufficiently anal retentive lawyer might tell you that
it does create a new derivative work, and thus, you should update the
copyright notice.
However, the worst that happens if you get this wrong on the side of
not a late enough date is that the protection date is calculated from
the earlier date. So you'd lose a year of copyright protection on the
protectible part of that derivative version (derivative copyrights
cover mainly the new work added to the derivative). None of us will
be alive when this code comes out of copyright in any case (in fact,
your kids will probably be dead as well), so it's probably not worth
worrying about in this case, unless someone official tells you to do
it.
:).
BTW, the answer to every legal question ever is "it depends".
--Dan
Thoughts?
Andrew