Python 3.0

What’s coming in Python 3000 (a.k.a. Python 3.0)?

First of all, Python 3000 will be out sometime in 2008; large parts of it have already been implemented. It is simply an increment on the current code base.

The biggest point is that Python 3000 will break backwards compatibility, abruptly. This is very unusual for Python, but is necessary to get rid of some old cruft. In general, Python has been very good about updating slowly and incrementally without breaking backwards compatibility very much; this approach is being abandoned for the jump from 2.x to 3.x.

However, the actual impact of this is likely to be small. There will be a few expressions that no longer work – for example, ‘dict.has_key’ is being removed, because you can just do ‘if key in dict’ – but a lot of the changes are behind the scenes, e.g. functions that currently return lists will return iterators (dict.iterkeys -> dict.keys).

The biggest impact on this audience (scientists & numerical people) is probably that (in Python 3.0) 6 / 5 will no longer be 0, and <> is being removed.

Where lots of existing code must be made Python 3k compatible, you will be able to use an automated conversion tool. This should “just work” except for cases where there is ambiguity in intent.

The most depressing aspect of Py3k (for me) is that the stdlib is not being reorganized, but this does mean that most existing modules will still be in the same place!

See David Mertz’s blog for his summary of the changes.