Hi Simon,
Simon Josefsson via Gnulib discussion list <bug-gnulib@gnu.org> writes:
> My impression is that even maintaining an accurate list of contributors
> to thank in a THANKS file is not particular useful any more. The git
> log is a more accurate source of who did what (and consequently should
> be thanked), possibly followed by ChangeLog in the (increasingly rare)
> cases where those aren't generated from the git log. I would prefer to
> use the THANKS files for acknowledging contributions from people or
> organizations who helped in some other way (e.g., hosting, websites,
> things that inspired the project) that wouldn't be clear from the git
> log.
>
> Even the AUTHORS file is becoming a bit less useful than it was. The
> authorship should also follow from the git log. Perhaps the more useful
> thing to record is current MAINTAINER (aka code owner) of particular
> source files and/or the entire project, which is often different from
> the initial set of authors.
>
> So I find e-mail addresses in any these files just another thing that
> gets outdated quickly. But going through these files cleaning them up
> is a lot of tedious work.
I generally agree that the 'git log' is more useful when trying to
figure out who did what, etc. But the point of THANKS is more for
distributed tarballs, in my opinion. For example on my Fedora machine:
$ find /usr/share -name 'THANKS' 2>/dev/null | wc -l
104
I think Debian gzip's them, but it has been a while.
Anyways, I like Coreutils THANKS.in file which uses the names contained
in it, 'git log' and the thanks-gen script to generate a THANKS file. I
think all (?) of the packages Jim maintains use it.
Maybe it would be useful to write a script to generate a THANKS file in
Gnulib, inspired by that.
Collin
