Re: Summary : ocaml, QPL and the DFSG.
On Wed, Jul 21, 2004 at 10:58:39AM +0100, Matthew Garrett wrote:
> Glenn Maynard <firstname.lastname@example.org> wrote:
> >Of course, XXX = "you must distribute source, too" is also a restriction.
> >Again, guidelines. (If the complaint is that these guidelines can't be
> >used without interaction with Debian and having the same result, then it's
> >just a complaint that they're guidelines--this can't be "fixed" without
> >turning it into something other than guidelines.)
> I would argue against any assertion that there's strong consensus that
> "distribute to upstream authors" is a "worse" restriction than
> "distribute source too".
I'll certainly throw my hat in in favour of "to upstream" being worse than
"source if binaries". Firstly, there's an "advancing freedom" argument --
ensuring recipients have source code (if they want it) has a great practical
advantage to freedom. I hope you agree with that (if not, we have more
fundamental disagreements than this small matter).
Next, there's the issue of cost -- presumably it is of trivial cost (or even
profitable) to me to distribute to my recipient, because otherwise I
wouldn't be doing it. It's unlikely that distributing source alongside the
binaries will significantly increase that cost -- and the GPL (the most
common example of this form of distribution) specifically allows the
recouping of distribution costs for source. However, it may not be a
trivial cost to distribute changes back to the original author -- in cases
previously hypothesised, it may even be illegal. It is also unlikely to be
trivial to determine what cost I may incur in sending the changes back
upstream at the time I decide to exercise my granted permissions.
Although it's not terminal to the point at hand, these "must send back to
the author" clauses have tended to be poorly written, having no time limit
or other effective means of limiting my exposure, so it is even harder for
me to determine any cost I may incur as a result of complying with my
obligations under the licence.
Finally, there is the matter of choice. I can choose who I distribute my
modified version to, and hence who receives the source. I cannot choose to
send my modifications upstream -- I am compelled to if I wish to exercise my
granted permissions. You may argue that I can avoid sending changes
upstream by not making changes, but that's a bollocks argument -- if I
cannot exercise the rights guaranteed to be available by the DFSG for a free
licence, then that licence is not free.