ACCEPTED: PEP 285

[Tim Peters]

[James Logajan]
> ...
> Seriously Guido, how often will I need to check back to www.python.org
> to keep up with the changes you are making?

It's unclear why you would ever need to, if you want to stick with an older
version, unless you wanted to pick up a pure bugfix release (a 3-digit
release, like next week's upcoming 2.2.1 and 2.1.3, being, respectively, the
first pure bugfix release in the 2.2 series, and the third in the 2.1
series).

> Do you have a schedule or some constraints in place on how fast you are
> releasing major versions?

Major releases change the first digit; there have been only three such in
all of Python history, counting 0.9.6 and 1.0 and two major releases.
Python 3.0 is at least years away.  The goal between minor releases (bumping
the 2nd digit) is 6 months.  This is what Guido strived for before and after
1.5.2.  1.5.2 was an anomaly.  More about the release philosophy can be
found in PEP 6:

    http://www.python.org/peps/pep-0006.html

> If you could please please please slow your major release rate down,
> and  make that an "official" policy

Why do you feel you can't slow down your adoption of new releases?  We'll
(PythonLabs) package and distribute bugfix releases in a series for as long
as people in the community care enough about a Python version to produce
bugfixes releases for it (speaking of which, you can think Michael Hudson
for 2.2.1 next week, and Anthony Baxter for 2.1.3 -- although I'll be
suprised if you do <wink>).

> I would reconsider my departure from Pythondom. It would be nice to see
> major releases at a _minimum_ of 18 to 24 months apart.

That's much faster than major releases happen now, as Python uses the word.
It's much slower than minor releases happen now.  If you only care to
upgrade every 2 or 3 years (whatever), you're not required to pick up
releases any faster than that.  It's not like you paid us for a subscription
and are going to get cheated out of your money's worth if you refuse to
accept delivery <wink>.

> And frankly I would think the law of diminishing returns has to kick in
> at some point; even if you disregard my request I would think it would
> become progressively more difficult for you to justify new releases all
> that often.

It's actually progressively easier:  Python is in a high growth phase, and
countless people want countless more things from it.  There is strong
sentiment for scaling way back on changes to the language proper, but that
leaves person-centuries of desirable work in improving the libraries and
implementation.

> When I see that happening (maybe in a few years?) then I would consider
> rejoining the ranks of Python programmers.

When the release rate slows to what you'd like to see, I'd take it as a
strong indicator that Python may be entering its death phase.  Indeed, I
took the long quiet after 1.5.2 as a sign that Python was in trouble, and it
turned out it was (note that Guido left his then-employer to get on with
2.0; apply some imagination).  It's not in trouble now, and frequent
releases are a sign of health for an open source project of this nature.