Do we still need weak copyleft? (was Re: Exceptions to copyleft-next)

[Bradley M. Kühn]

Daniel P. Berrangé wrote:
> I've created and maintained several C libraries under the LGPL,
> with the rationale that I want them used as widely as possible.

I think what I'm really questioning is whether this is a model that
exists in the modern world of software.

Are *new* LGPLv2.1'd libraries likely to get adopted (and funded) by
proprietary vendors?  Do we have any ideas for how to make that
happen?

I guess it's also a question of priorities.  Making a weak copyleft
is actually a lot harder than it looks (unless you want it to be
pathetically weak like the Classpath Exception — and I say that as
one of its co-authors!).

I'm doing a panel discussion next month at FOSSY 2026 about the GCC
RTLE.  I co-wrote that too (and Richard Fontana was involved for part
its drafting, too).  GCC RTLE works — more or less — but it's way too
clever for its own good.  And, despite best efforts, FSF couldn't
convince Apple that GCC RTLE suited their licensing needs so they'd
keep contributing to GCC.  Apple instead literally used the GCC RTLE
as part of its internal justification to finally go all-in on LLVM.

Like, I'm sort-of proud of the GCC RTLE because of its cleverness,
but that experience really taught me that Additional Permissions (be
they formalized in an LGPL-like thing or ad-hoc like the Classpath
Exception) just generate more questions than answers for those who
want to build something proprietary and incorporate the Work.

The magic of the LGPLv2.1, frankly, is the fact that very few really
have a deep understanding of LGPLv2.1, but everyone has some
assumptions about what its requirements are.  Literally, like two
months ago, I realized something really complicated and important
about the LGPLv2.1 that I had not realized after *30 years* of
reading it all the way through at least once a month (on average).

In theory, weak copyleft is great.  In practice, weak copylefts
quickly become impenetrable documents that folks squint at and
pretend they're some beautiful middle ground.

Anyway, I'm curious if Richard agrees, but it seems to me there is
probably zero chance we'll try to develop a weak copyleft alongside
copyleft-next 1.0.  If nothing else, I've hopefully proven in this
thread that creating a *good* weak copyleft is at least order of
magnitude harder than creating a good non-weak (to use Richard's
phrase) copyleft.  So, maybe we should just close this thread and
revive it immediately after copyleft-next 1.0 is released. 😀
--
     -- bkühn — he/them — On Fediverse (via Mastodon): https://fedi.copyleft.org/@bkuhn