Bugfix releases (RE: profiler results for __getattr__, am I reading this correctly ? )

[Steve Holden]

"Anthony Baxter" <anthony at interlink.com.au> wrote ...
> >>> "Tim Peters" wrote
> > The bugfix release process is the topic of Aahz's PEP 6, who may have
> > believed that titling it "Bug Fix Releases" was enough clue <wink>:
>
> Ah, that cunning "hidden in plain sight" approach. Who'd have thought.
>
Gets *me* every time...

> It still doesn't cover the process of deciding what would go in the
> patch release... I know that I'd be happier upgrading to 2.1.2 than
> 2.2 in the next couple of months - and given the peasants-are-revolting
> sentiment of c.l.py at the moment, I doubt I'm the only one...
>
It's only the fatc that they're revolting that identifies them as peasants.
Seriously, though, there are many good reasons why you might not want to
upgrade, including having a large and varied hardware base and wanting to
keep all Pythons roughly in step. Hence, lowest common denominator, usually
what you started with on your first machine (though personally I removed
1.5.2 from my last system three or four moths ago, I'm now thinking of
re0installing it on a test machine for compatibilty testing).

> How would this sound as a first cut at a what-goes-in-and-what-stays:
>
>    a) bugfixes.

Kind of essential in a bugfix release.

>    b) no new modules, no new packages.

Definitely.

>    c) no new methods on classes, unless they're bugfixes.

Indeedy.

>    d) if the bug fix relies on some 2.2-ism, then it's either rewritten,
>       or ignored and relnoted
>
Yes, but are there any such?

This all sounds about right. So, how does it feel to be the new patch czar?

regards
 Steve
--
http://www.holdenweb.com/