Do you QA your Python? Was: 2.1 vs. 2.2

[Bengt Richter]

On 13 Apr 2002 21:35:24 -0700, Paul Rubin <phr-n2002a at nightsong.com> wrote:

>Tim Peters <tim.one at comcast.net> writes:
>> It's in the nature of open source development that volunteers scratch their
>> own itches, then share what they've done because it does them good too to
>> have their work widely adopted.  If you can ride that curve, you can get an
>> enormous amount of high-quality work for free.  If not, you'll have to pay
>> for it or go without.
>
>A lot of stuff gets done because it needs to get done, rather than
>scratch an itch.  
>
What's a need but a very insistent itch? ;-)

>> A crushing amount of debate over "stability" has gone by on Python-Dev the
>> last week, and the one thing that strikes me over and over is that, with
>> just a few minor exceptions, nobody volunteers to *do* anything except tell
>> other people what to do.  What are you willing to do to make your desires a
>> reality?  If it doesn't involve contributing work, time or money, you just
>> want a free ride.  But nobody owes anyone a free ride, and open source can't
>> change that fact of life either.
>
>In that particular debate, a lot of people were asking for certain new
>features to be LEFT OUT of Python.  I don't understand how someone can
>volunteer to contribute work, time, or money toward leaving something out.

Maintaining older baseline releases, where the features are already left out?
I don't think that work is zero, so should the BPWIs[1] be doing that, and who should say?

The trick is what happens when various versions are adopted out in the real world
without asking those whose scripts are expected to continue to run.
Can one distribute a script with first line #!/usr/bin/python21 and expect the
link to be there wherever python21 (or fully compatible other version) is available?

Otherwise new versions will potentially make extra work for me, which I'd rather
volunteer in other ways ;-)

Apparently 1.5.2 and 2.1.x and 2.2.x are semi-official baseline versions. Is it/could it
be official policy that #!/usr/bin/python(15|21|22) links should exist if a site
has the identified version? If /usr/bin/python is going to be redirected every six months
forever, I think I'd rather indicate a specific version (and it wouldn't make me unhappy to).

OTOH, if someone is motivated to consolidate all so they can vector all to the same latest
interpreter on their site, there's nothing to prevent it. It's their time and responsibility
fixing breakage and validating. If they share some validated upgrades, no doubt there will be
appreciation, but (ideally) no one's dragged into any work they didn't plan on.

[1] Benevolent Programmers While Interested ;-)

Regards,
Bengt Richter