conditional expressions (RE: Loop-and-a-half (Re: Curious assignment behaviour))

[Guido van Rossum]

Donn Cave <donn at u.washington.edu> writes:

> Right, I think it's official that your (our) vote doesn't count, but
> we do count as users, and like anything else, if we just take what
> we're handed, we get what we deserve.
> 
> I don't know how I'm going to feel about all this, down the road, but
> there's a chance I won't be on board, in the sense that I will not be
> interested in upgrading.  That's the vote I get.
> 
> It won't mean much if it's just me.  I don't think it would take all
> that many, though, to make a difference.  I don't expect GvR et al. to
> undo all the changes they have been checking in, or even to necessarily
> take notice of whether they're losing some old timers - from the sound
> of it, I'd expect more like "good riddance!"  But the big question for
> those of us who bail out is going to be, I think, whether we continue
> to invest in our old version of Python, or put Python development in
> legacy mode and start looking for a more attractive option.
> 
> Right now, that doesn't sound all that wonderful - far too radical,
> for the minor problem at hand, maybe.  But when the time comes,
> for each of us, it will be a lot easier for all of us if we have
> collectively already settled a "classic Python" version.  Which
> one I don't know, 2.1, 2.0, even 1.5.2.  If there's a solid,
> permanent constituency for that one specific version, then even
> outside that world, maybe a few people who write Python software
> will make a point to bear in mind what will work there.  If that
> happens, operating system vendors and others who can't support a
> constantly changing Python might be attracted.  If it worked, it
> could be step forward for Python.
> 
> The hard part is that we have to make up our minds to do this.  If
> everyone falls off separately, we'll be scattered all over, because
> it's hard to downgrade.  Everyone has to decide for themselves, but
> at any rate, please, please don't upgrade to 2.2 if you have doubts.
> Don't feel that you have to jump off while you still like what you're
> getting, but don't jump off after having upgraded to a version you
> don't really like!

Donn, I do care about you -- in particular because you've been a
valuable contributor over the years (that's about the only quality
measurement I have for community members :-).

I don't know what it takes to reconstitute your faith in Python 2.2.
There's one little point of light: conditional expressions are out.  I
couldn't find any examples in the standard library that would be made
to look better by using them. :-)

Regarding all the other changes (you're right, I won't back them out
:-), I believe that they are much more backwards compatible than you
fear.  Integer division won't become the default until 3.0, for
example.  Same thing for new-style classes: I'm committed to make
applications that use Jim Fulton's ExtensionClass or otherwise use the
Don Beaudry hook work the same way as they did in 1.5.2 and 2.1.1, and
I've pretty much achieved this in 2.2a4 already.  New-style classes
will be dynamic by default in 2.2b1 and their instances will allow
assignment to self.__class__.

What else?  2.2 has lots of improvements that you'll want and that
don't affect the language syntax at all -- better portability, better
and automatically configured large file support, etc.

--Guido van Rossum (home page: http://www.python.org/~guido/)