Re: d/copyright years when upstream does not use years?

Hi!

On Fri, 2026-08-28 at 09:43:59 -0700, Russ Allbery wrote:
> For the purposes of determining when code goes out of copyright, I would
> think the version control repository would be even better. The dates feel
> like a pre-version-control fallback.

This might(?) assume that all VCS users properly attribute authorship
via the native matadata (such as --author in git), instead of as a
comment in the commit message, or on a source file. From what I've seen
many people seem to either not know about such options or not care to
do that.

> That fallback is not without value; there are a lot of situations, such as
> Debian packaging, that often discards the original version control
> repository when producing widely distributed artifacts. But a preserved
> original version control repository or some timestamp-preserving
> conversion of it seems strictly better than the copyright notices for this
> purpose, since it attaches a date to every hunk of code with context and
> isn't constrainted to the granularity of a copyright notice.

Including old projects with long histories spanning two or more VCS
that did not have a way to represent the difference between the
committer and the author (such as CVS, or svn).

So, even when you have the VCS, it might not be as convenient as just
checking the comment header for a file (if the copyright statements are
there :).

Thanks,
Guillem