On 29/05/2020 12:26 pm, Chris Angelico wrote:
On Fri, May 29, 2020 at 11:57 AM Mike Dewhirst <mi...@dewhirst.com.au> wrote:
I am an example
I installed all the Pythons on my Windows 10 dev machine (locked into
Windows by having clients) but I'm also locked into Python 3.6.9 on my
Ubuntu 18.04 production machines.
Be careful of assuming too much here. The general policy with most
Linux distributions is to number the version according to the oldest
component in it, but they are free to backport whatever changes they
choose. I have Python 3.4.4 and 3.5.3 on this system (Debian Stretch),
but the 3.5 is numbered "3.5.3-1+deb9u1" which implies some collection
of additional patches. (I'd have to dig deep in the changelogs if I
cared exactly *which* patches.)
After chasing down an obscure problem I decided to go back to Py36 on
Windows to be using the same versions in dev as in prd. I couldn't find
an installer on python.org so I retrieved one (3.6.5) from my archives.
That'd be the same feature version, but depending exactly what the
problem is, there might not be *any* Windows build that exactly
corresponds.
If I was asked to suggest a guide for which versions ought to get a
Windows binary I would look at the most popular LTS *nix distros and
keep Windows binaries in step just to support people like me who cannot
live with too much Windows clutter. Think of it as deeply humanitarian
generosity.
Even if that were possible, who are you asking to do this? Whose time
is going to be put into finessing every point release to make sure
it's still buildable on Windows?
I was careful to not ask anyone to do anything. "If I was asked ..."
Making sure something is buildable when it involves compiling on Windows
is well beyond my pay-grade nowadays. I know just how hard it is.
However, I would say (and I've said it before) Christoph Gohlke deserves
a Nobel prize for his abovementioned deeply humanitarian generosity.
Without him (and Mark Hammond) there would be very little Python dev
done on Windows - if any!
The only reason I took up Python (after being burned by proprietary tool
makers) was because it was promised to run on pretty much everything.
Especially Apache platforms - which also run on Windows.
"LTS" doesn't mean anything since
basically EVERY version of Python is in an LTS of Red Hat (they're
probably still shipping Python 2.3 somewhere).
Honestly, if you let it, Windows just absolutely knows what you really
meant despite what you tell it. It is a necessary evil when your clients
use it.
If your clients use it,
They don't use Python. They use Windows and I need to use Windows to be
able to support them in what they do. I install Apache on Windows if I
need to.
It would be handy if Microsoft delivered Windows with Python installed
but what can you expect from them if they can't see demand for it.
are you going to pay someone to build your
installers? And if you aren't going to pay, who is?
I agree with your sentiment. But without specific funding I can't afford
to pay anyone. I'll choose different solutions. But if I needed it badly
enough I'd find the brainspace and do it myself.
Find someone who'd be willing to maintain Windows binaries for you,
and see what they'd charge you for that. Then judge that against the
cost of dropping support for Python 3.6 and requiring your users to
upgrade to a fully-supported version.
Chris, you are preaching to the converted. You won't get a
counter-argument from me. I piped up in the first place because I saw
the OP in trouble and felt it was necessary to say something. I thought
my scenario might be interesting. As I said earlier it isn't critical
for me.
There are lot of Python people working on Windows and they do so because
they have no choice. Windows is simply the dominant platform. That's
where most clients live.
All I'm saying (now) is I really hope PSF is supporting Christoph and
Mark because they make it possible to use Python on Windows.
Cheers
Mike
ChrisA
--
https://mail.python.org/mailman/listinfo/python-list
