[Python-Dev] 3.2.0

[Nick Coghlan]

On Fri, Feb 18, 2011 at 3:19 AM, Terry Reedy <tjreedy at udel.edu> wrote:
> Actually, to me, the confusion is slightly worse, and the reason to change
> slightly stronger, than I initially explained. Python x.y is a version of
> the *language*. CPython x.y.z is an occasional marked release of an
> *implementation*.
>
> For instance, Python 3.2 is a version of the language and stdlib. It has
> been pretty well defined since new features were prohibited.
>
> The 3.2 docs are the specification of Python 3.2 (with a few
> CPython-specific notes). The 3.2 docs will be continuously upgraded as
> deficiencies are noted and fixed. As I understand it, all patches are
> expected to leave the docs in an improved and buildable state, so that
> updates can be built and uploaded to the site frequently (daily?).
>
> CPython 3.2.0 will be the first 'production' release of the CPython
> implementation of Python 3.2. It will be one in a series of approximation of
> an ideal bug-free 'CPython 3.2'. Some have already been released, more will
> come. Like the docs, the concrete CPython 3.2 codebase will also be
> continuously upgraded. For various reasons, it will probably not always be
> buildable on all platforms and not always be production ready. For practical
> reasons, marked releases will be spaced some months apart.
>
> So, for me, Python 3.2 is a now theoritically fixed version of the language.
> The Python 3.2 docs document that version, but will be upgraded as mistaked,
> ambiguities, and omissions are found. The CPython 3.2 codebase is an
> evolving approximation of an ideal bug-free CPython 3.2 (that will never be
> reached). And CPython 3.2.0 is an early snapshot release of that evolving
> codebase.

I actually agree with this viewpoint, and think it would definitely be
a good way to go for 3.3.0.

For the 3.2 series, I think living with the ambiguity for another 6
months or so (however long it is until 3.2.1 is released) is the
better choice. There are enough parts of the release process that
involve the version number that we *really* shouldn't be messing with
it during the RC phase.

Cheers,
Nick.

-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia