Date: Tue, 04 Dec 2018 11:46:00 +0000 From: bugzilla-noreply@freebsd.org To: ports-bugs@FreeBSD.org Subject: [Bug 233770] [exp-run] make lang/python37 default python3 Message-ID: <bug-233770-7788-OD1CDhURty@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-233770-7788@https.bugs.freebsd.org/bugzilla/> References: <bug-233770-7788@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D233770 Kubilay Kocak <koobs@FreeBSD.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags| |exp-run? Status|New |Open Priority|--- |Normal Keywords| |needs-qa --- Comment #1 from Kubilay Kocak <koobs@FreeBSD.org> --- I'm all for moving as quickly to using the latest and greatest in the Python ecosystem. However, we need to keep in mind that: a) The vast majority of Python ports, currently use loose and implicit rath= er than precise and explicit version specifiers. That is, either;=20 1) Under-specify/declare their version support, with a major class example being "X.Y+", without an upper bound, or 2) Don't specify version support at all (bare USES=3Dpython), used to imprecisely mean something like 'all versions' or 'python 2 and 3'. b) Python package support for the latest (usually 1-2) and greatest version= s is a fairly slow moving target. It takes time and manual interventions upstrea= m, for those later versions to enter the CI ecosystem/configuration into a 'te= sted by default' status. c) With the vast majority of Python ports sitting without test suites, an exp-run can only tell us a limited amount about potential issues, and in particularly, not at run-time, in order to give us a good answer to 'is it a good time to switch'. An exp-run should be a first step, and not the only step/data point, to deciding to move forwards on later versions being a default, no matter how great it would be to do so. We should consider within that decision other issues such as, but not limit= ed to: - Support periods (if they differ) between releases - Time / number of releases since releases - How to improve Python port version specification / explicitness to make it easier to move default versions without introducing unknown/untestable regressions (until after the fact). The question I believe we actually need to pose and ultimately answer (outs= ide of the scope of this exp-run/issue) is, what (else) do we need to do to make moving default versions forward a trivial/non-event. A large part of that question is whether we take an "exclusive allow" vs "exclusive deny" approach to version support in Python ports. Right now we'= re in 'exclusive deny mode' where most ports are loosely or under specified (incorrectly), until bugs/issues are reported. @Ruslan Maybe its worth a few of us (python) getting together on IRC to fle= sh out a plan for this. It'd be great to make it easier for these things to mo= ve forward faster in the future too. Think: python default =3D 3.x, which is a superset of this problem. --=20 You are receiving this mail because: You are on the CC list for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-233770-7788-OD1CDhURty>