I second that. Unless someone presents us with a pressing case of a large Python 2-based library unportable to Python 3 (and therefore already in a dead-end), our limited workforce is better employed to maintain our Python 3 Sage.
In other words, "if you need a Python 2-specific fix, speak *now* or forever hold your peace..." Corollary : we should update the places where we published our plans for Python 2 compatibility and point to the present thread, which can be summarized by "The Sagemath developers have decided not to maintain Python 2 compatibility beyond Sagemath 9.0." as an acceptable first-order approximation. I agree that a Python 2-compatible maintenance branch *would* be a Good Thing (TM) to have. But I doubt that our present workforce is enough to maintain it. A french proverb says that "L'enfer est pavé de bonnes intentions" (Hell is paved with good intents) ; I think that we have such a case... Now, for the cleanup of Python 2-specific idioms, I suspect that this will take a long time. And defining the set of Python 2 expressions that would be advantageously rewritten with (different) Pyhon 3 idioms is by no way an easy task. I would suggest to keep it as an informal goal, and to treat it incrementally. HTH, Le mardi 21 janvier 2020 17:19:33 UTC+1, Timo Kaufmann a écrit : > > > > Am Freitag, 17. Januar 2020 16:07:08 UTC+1 schrieb E. Madison Bray: >> >> On Wed, Jan 15, 2020 at 2:37 AM Matthias Koeppe >> <matthia...@gmail.com> wrote: >> > >> > On Tuesday, January 14, 2020 at 12:01:42 PM UTC-5, Frédéric Chapoton >> wrote: >> >> >> >> So here is my proposal. >> >> >> >> * Starting from now, we allow ourselves to move on, using 9.1 betas >> and further releases for external python3 updates, including switch to >> ipython7, which seems to me the most urgent matter. But we also do not >> introduce python3-only code in our own code base if we can avoid it. >> >> I'm not exactly sure what you mean by this, but it might be agreeable >> to me: I'm completely okay with adding Python 3 only dependencies and >> even *code/features* so long as it's done without breaking >> backwards-compatibility. When it comes to sage-the-distribution >> (which I sense is where most of this friction is coming from, and yet >> another reason to better separate sage from the >> sage-the-distribution), if there are dependencies you want to update >> that are Python 3-only I'd say go for it, but make it a separate SPKG >> so that previous versions of the dependency can still work on Python 2 >> builds. >> > > As others have said already, supporting both python2 and python3 > dependencies is much more work than only supporting python2 or only > supporting python3 versions. > > I think dropping pyhon2 support with the next release remains the best > option. Even better of course would be to keep the option for bugfix > releases for 8.9 / 9.0. That would be a nice thing to have in general, but > we could also just make an exception for this one release. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/459616b9-b12b-4588-b684-7833536e2db9%40googlegroups.com.
