On Tue, May 21, 2013 at 10:26:37AM -0700, Luis R. Rodriguez wrote:
So while I do
understand the ideas of evolving copyleft to a point perhaps we've
never been at I think it is important to not evolve it too fast, or if
we do at least allow for advancing the older licenses. So for example
as with Linux with stable releases in which we have on linux-stable:
* linux-3.8.y
* linux-3.9.y
* linux-3.10.y
I'm not sure how this model can be carried forward to licenses, since
there are plenty of projects (not to mention core code in the Linux
kernel) which is two decades old at this point.
With the Linux kernel, after 5-10 years, we can be pretty certain that
older kernels simply won't boot on modern hardware, and the old
hardware has probably been decomissioned already. But that's not true
with specific lines of code. So even if we were to migrate forward as
you propose:
* copyleft-0.1.y
* copyleft-0.2.y
* copyleft-0.3.y
The cycle time is probably not going to be measured in months, as it
is in with software releases, but more like decades. Put another
away, eventually at some point the Linux Kernel will be replaced
whole-sale by some new OS --- I will be (pleasantly) surprised if we
are still using Linux 50 years from now --- but so maybe then, if we
assume that whatever new OS comes around doesn't use a single line of
code from Linux or any other GPLv2 project, maybe we could forsee some
point, half a century from now or later, when GPLv2-only code finally
"dies".
Also, with linux 3.8.y, 3.9.y, 3.10.y, etc., it's possible to forward
port and backwards port patches between the various different kernel
versions. One of the problems with assuming different license
versions is that you may only be able to port code in one direction,
or in the case of GPLv2-only and GPLv3-only code, the code can't be
ported in either direction.
And in between GPL-2.y and GPL-3.y I see a huge change, and in
between
GPL and AGPL I also see a huge change. The evolutions between them and
the changes within should be considered gradually. We are at a point
today where for example the economics simply don't even exist to
enable AGPL to fruition in the market due to the lack of an
appreciation for such licenses. This is why I'm concerned with the
rapid evolution and expectations of embracing advances in copyleft.
It's not obvious to me what the business model would be that would
allow startups or companies to use the AGPL without negatively
impacting their businesses (to be at least neutral, let alone actually
being a net positive for them). Regardless, the big problem is how do
we square the desire for code sharing between projects with the desire
for "rapid evolution" of licenses.
- Ted