Hi Soren,

On 06/02/2025 22:04, Soren Stoutner wrote:
On Thursday, February 6, 2025 2:33:13 PM MST Ahmad Khalifa wrote:
 > >  > > Files: inc/gmp.bi
 > >  > >
 > >  > >  bootstrap/inc/gmp.bi
 > >  > >
 > >  > > Copyright: The FreeBASIC Development Team
 > >  > >
 > >  > >            1991-2014 Free Software Foundation, Inc.
 > >  > >
 > >  > > License: GPL-2+, and GPL-2+ or LGPL-3+
 > >  >
> >  > And this is one of the occurrences where the original was A OR B, so now
 > >  > it says "A AND (A OR B)", which again looks silly.
 > >
 > > I recently did something like this:
 > >
> > https://salsa.debian.org/debian/maildrop/-/blob/master/debian/ copyright?
 > > ref_type=heads#L68
 >
 > Oh this works and lintian doesn't complain about parenthesis!
 > Section 7.2 in that copyright format has this obscure comma based
 > format. More natural language, but slightly ambiguous.
 >
 > But are you also saying to keep the AND/OR and not squash them down?

The LGPL-3+ cannot be subsumed into the GPL-2+ without turning it into the GPL-3+.  So, I think the only way to accurately state what happened is:

License:  GPL-2+ and (GPL-2+ or LGPL-3+)

This says that the original file was available under the GPL-2+ or the LGPL-3+.  The FreeBASIC Development Team decided to utilize the file under the GPL-2+ option, and then licensed their code translation under the GPL-2+.  This means that the current version of the file is only available under the GPL-2+.

Alternatively, you could do:

License:  GPL-2+
Comment:
 Free Software Foundation licensed the original file was under the GPL-2+ or
  the LGPL-3+.  The FreeBASIC Development Team decided to utilize the file
 under the GPL-2+ option, and then licensed their code translation under the
  GPL-2+.

Understood, no squashing due to the compatibility this time, but in other cases it's ok. Going to go back to the compatibility matrix for a bit before making changes.

 > If I may debate a little here...
 > You're quite right in that the accurate picture is to group the files
 > according to their copyrights and licenses. And even to separate out the
 > HPND licenses into their verbatim copyright holder versions: HPND-DEC,
 > HPND-Keith, HPND-SGI, HPND-HP, ...
 >
 > But in practice, what's wrong with adding licenses, even if they're
 > superfluous?
 > If the goal here is to protect debian from any incorrect usage due to
 > debian's declaration of copyright on distributed software, isn't it safe
 > to pile on more licenses? Similar to how we lump up several copyright
 > holders over multiple files.
 > It only makes it "inconvenient" to use the files without reading their
 > individual license in the rare case that someone wants to fish out a
 > single file and lump it with another non-free license.
 >
 > In other words, isn't it safer to add an extra license to a whole
 > directory than to forget a license on a single file?
 >
 > Of course, this is partially motivated by the objective to reduce the
 > maintenance effort here, but can't help wonder why no other distribution
 > system has this level of granularity and still thrives (fedora, flatpak,
 > ubuntu, ...).

If that were the case, then each debian/copyright file would just contain one stanza.

Files: *
Copyright: All the copyright statements in the package.
License:  All the licenses in the package.

You propose that like it's a bad thing, but I would find it amazing if it's possible. It's fair to the authors who get recognised similar to an AUTHORS/CONTRIBUTORS file, and it's a nice heads up to anyone who wants to use the source of what licenses may be involved. I could have saved 2 weeks fiddling with coreboot's copyright file last year if this was an accepted option.

This is counter to the design of DEP-5.  If you feel strongly about this, you are welcome to propose an update to DEP-5, but I think it is accurate to say that Debian’s current practice is that the license listed under a files stanza should apply equally to each file in that stanza.

I feel strongly about it, but this is beyond my pay grade. (and also having seen some previous discussions on DEP-5, I know there are other strong opinions there too)


--
Regards,
Ahmad

Reply via email to