Re: switching to Python 3.x

[David Kastrup]

Han-Wen Nienhuys <address@hidden> writes:

> On Sun, Jan 26, 2020 at 3:04 PM Jonas Hahnfeld <address@hidden> wrote:
>> > > > So far, I've only received a single (positive) response off-list and a
>> > > > bit of feedback on the posted patches. What do others think?
>> > > > To make this explicit: The proposal is to drop support for Python 2
>> > > > (now EOL), requiring everyone wishing to build LilyPond 'master' to
>> > > > have an appropriate version of Python 3 available. This should be
>> > > > sufficiently easy (see above), but I'd like to have consensus on this.
>> > >
>> > > When we switch over GUB, we also need to switch over the 2.20 branch.
>> > > It's not just master that is affected.
>> >
>> > But we only need to switch GUB to py3 when we package 2.21 with GUB.
>> > We could kick this problem down the road until we really need a GUB
>> > build of 2.21, which may well be a few months in the future, or never
>> > if we decide to move to some other packaging mechanism (cf. the
>> > discussion of Docker/Windows)
>> >
>> > I don't think it's reasonable for us to ask Jonas to also package
>> > python3 for GUB as a precondition to getting his patches in.
>>
>> I'm going to start by quoting myself:
>> > > > > On the GUB side, I already added a spec for Python 3.7.4 (also for
>> > > > > Windows via binary packages) and this worked successfully in 
>> > > > > September.
>> So even if you don't think it's needed, I already did. Even though I
>> would fully support switching away from GUB!
>>
>
> Kudos for biting through GUB.
>
>> What David is concerned about (as far as I understand) is that we need
>> to modify the spec for LilyPond to require the new python3 package as a
>> dependency. This will (obviously) not work for packaging 2.20.
>
> Fair enough, but that would only be a problem if we ever have to
> produce a 2.20.1 . We could delay 2.21.0 for a while.

I prefer pushing out 2.21.0 right after 2.20.0.  It's been too long in
the making by far already.  And while the 2.20 release is not quite
within the time frame I aimed for a week ago (contact me for details), I
don't expect the additional delay to be able to push 2.20 beyond
February.

> If we get lucky, we never have to produce a 2.20.1. If we do, we might
> have to backport the py3 patch.

2.20 is in maintenance mode for as long as 2.22 is not out yet.
Considering the amount of stuff stacking up on the 2.22 slate right now
regarding platform support alone, that is likely going to be a while.

I don't think that Python3 will port to PowerPC, so a backport of
Py3-only code would entail cutting its support in the middle of the 2.20
lifetime.  To be honest, I already had suggested cutting it before 2.20
but was met with resistance.

-- 
David Kastrup