Re: Python 3.6 in stretch
On Mon, Jan 9, 2017 at 1:54 PM, Ben Finney <bignose@debian.org> wrote:
> Adam Borowski <kilobyte@angband.pl> writes:
>
>> On Tue, Jan 10, 2017 at 05:35:34AM +1100, Ben Finney wrote:
>> > Andrey Rahmatullin <wrar@debian.org> writes:
>> >
>> > > On Sun, Jan 08, 2017 at 06:55:45PM +0100, Galbo Branbert wrote:
>> > > > Thanks for the info, didn't know that the transition freeze was
>> > > > actually the version freeze for minor versions of Python.
>> > > A minor version upgrade would be 3.5.3 -> 3.5.4. 3.5 -> 3.6 is a
>> > > lot of changes.
>> >
>> > Galbo is referring correctly to the minor version, as specified in
>> > <URL:https://www.python.org/dev/peps/pep-0440/> and Semantic Versioning
>> > <URL:http://semver.org/>.
>> > […]
>>
>> Not every project uses semver.
>
> We're not talking about “every project”. We are talking specifically
> about Python, where “minor version” has the meaning Gablo used.
I agree, Python intends to use semver. However, in practice, they do
not. If 3.5 to 3.6 was a typical "minor version," our expectation
would be that the update comes with security updates and bug fixes
(not feature changes). Python uses the microversion position for this.
More importantly... is this really a point that matters?
Is not the point that matters the fact that 3.5 to 3.6 introduced
significant changes? As such, it seems the argument about whether this
is a minor version bump, a significant one, or whatever we wanna call
it are moot. The version change /does/ introduce significant changes
and this carries substantial risk to even consider rolling at this
point.
Remember, we have a long freeze, not so we can continue to sneak in
cool features, but rather so we can ensure a high quality release.
Reply to: