On Thu, Jun 27, 2019 at 4:58 PM Ben Cotton <bcotton(a)redhat.com> wrote:
On Thu, Jun 27, 2019 at 2:52 PM Dan Book <grinnz(a)gmail.com> wrote:
>
> As an outsider to the Python community, not having any binary or package that
responds to the expected name "python" would be a disaster.
>
Can you expand on that? As I understand it, most things that are
calling for "python" now are expecting that to be "python2". So when
it becomes "python3", they'll break anyway. So why perpetuate a
pattern that's not future-proof (for some values of "proof")?
The majority of the Python community writes Python 3 code pointing to
an unversioned shebang, even when it's Python 3 only. This is because
in everything *except* Linux distributions, the unversioned name has
already switched over. And OpenMandriva made the switch before us, and
that was not a terribly painful change like it was for Arch many years
earlier.
In fact, I would argue that the only mistake was in RHEL, where RHEL 8
did not ship with the default unversioned names that would point to
the Python 3 variants.
--
真実はいつも一つ!/ Always, there's only one truth!