[Python-Dev] PEP 594: update 1
steve.dower at python.org
Thu May 23 12:57:57 EDT 2019
On 23May2019 0947, Anders Munch wrote:
> Fra: Paul Moore [mailto:p.f.moore at gmail.com]:
>>> A major version change serves as a heads up that something is going on and you need to check the consequences before upgrading.
>> Python's backward compatibility policy allows breaking changes between versions X.Y and X.Y+1 (with a suitable deprecation period). This proposal is no different.
> Except perhaps in scale. The same people that upgrade from 3.x to 3.x+1 without giving it a second thought, just to be on the latest version, will hesitate to go from 3.x to 4.y, because the major version change is a hint that they should be more careful. That means they're ready for it when they get the ModuleNotFoundError exception, instead of feeling ambushed.
> OK, it may be that this is not enough to warrant a 4.0 release, but I do think python-dev should get over its fear of major versions sometime. And that transitioning to a leaner standard library with some things moved to PyPI would not be a bad program statement for a Python 4.0.
We need to make it more clear somehow that Python uses
series.major.minor.micro versioning, not SemVer. I see this confusion
happen all the time.
Think of it as Python3 7.x moving to Python3 8.x, and evaluate how much
effort you should put into migration :)
That said, I'm totally in favour of Python 4.0 being the point where we
change expectations about what will be in the standard library (and also
C API, language/implementation semantics, etc.).
But until we have a clear vision statement that everyone (or enough) is
behind, there's nothing gained by prematurely causing that much disruption.
More information about the Python-Dev