Re: Python 2 fade-out

[Greg Troxel <gdt%lexort.com@localhost>]

Brian Buhrow <buhrow%nfbcal.org@localhost> writes:

> 	hello.  One of the more popular packages, I think, which relies on python2 is
> mail/mailman.  The version of mailman in the pkgsrc tree appears to be V2.1.35.  It doesn't
> look like anyone has created a package for mail/mailman3.  We should probably have a packaged
> version of mailman3 in the pkgsrc tree before depracating the v2 version of mailman and, thus,
> python2.7.

great to mention it, but I do not interpret the current discussion as
agreeing that we'll start deleting packages because they depend on
python2.

I see it as approving a change that we won't be building 27 versions of
packages if the package is silent about 27; getting a 27 version will
need a variable to say it works.   That's likely true already for most,
and easy to add.  This won't withdraw any end-user packages, but will
mean that people who expect py-foo to have a 27 version, because they
use it privately, will have to speak up or adapt.

I suspect mailman2 is going to be one of those things like scons that
will take a long time to reach non-use.

mailman doesn't build with the proposed change to drop 27 from the
default set.  I suspect that's just because py-cryptodome doesn't assert
27 support.

Do we want a new "this package supports 27" variable, vs having to set
ACCEPTED to an explicit set, sort of

PYTHON_27_ACCEPTABLE=	yes

with an "if set, add 27 in to the default set" stanza?