[License-discuss] popularity, usage, re-review of old licenses

Thorsten Glaser tg at mirbsd.de
Sat Jun 8 02:30:43 UTC 2019

Luis Villa dixit:

>But if a license can't be found in the tens of millions of projects on
>GitHub, plus Fedora/Debian[2], then for many (all?) policy purposes I think
>OSI should feel comfortable saying "this is unused, or close enough that we
>feel OK treating it as unused"...

It should most certainly not!

Not everything is on GitHub. Besides, it’s a proprietary platform,
which doesn’t even require things to be under OSS or otherwise
free-ish licences (or any at all other than the one to them from
the ToS), and, worse, whose ToS forbade some OSS from being uploaded
to them for a while.

GNU distributions like Debian and Fedora have very specific use
cases. You won’t find most Java™ packages there, for example,
and those new hipstar languages even less.

I also think you misunderstand what the Software Heritage
archives are about, but my point here is that, no matter how
many individual archives you include in your list-to-search,
you won’t be even close to anything acceptably representative.

Worse: this is especially bad for special-purpose licences or
those that have their user base mostly in a small, specialised
community (like the EPL in the Java™ ecosystem). If you’d just
look at the distributions you’d maybe see a dozen source packages
using them. Your analytics are biased *against* those you’re
looking for.

>[2] I suspect OSI should probably focus on *live* software in most policy

And no, this is also wrong. The right to fork doesn’t terminate
after a decade or even two (I have resurrected code from the
1970s due to licencing issues of newer code.) The point of going
through the annoying process of submitting a licence to OSI is
making it possible for, say, forks of such old code under those
licences to be allowed to be hosted at certain platforms (like
SourceForge, who demand(ed, last time I looked) an OSI-certified
licence) or for inclusion into some software where the contract
states such a requirement, etc.

PS: Besides, when was popularity ever a good indicator?
    Perhaps if it was, the world would run with IE6 still.
Solange man keine schmutzigen Tricks macht, und ich meine *wirklich*
schmutzige Tricks, wie bei einer doppelt verketteten Liste beide
Pointer XORen und in nur einem Word speichern, funktioniert Boehm ganz
hervorragend.		-- Andreas Bogk über boehm-gc in d.a.s.r

More information about the License-discuss mailing list