I've used Python 3.1 since it was released. I think most of the changes made in Py3k were good ones, so it seemed like the right thing to do. Now though, I'm sad to say I'm likely going to switch to Python 2.7. Ugh. Think of the version control diffs :(
Every single library I've gone out to use wasn't compatible, and likely won't be this year. I've ported a couple, which wasn't hard, but silly that I had to do it. I've worked around the needs for others, but it's just not a reasonable way to work. Plus, when I've sought help, everyone balked at me using 3.
SciPy (NumPy) was the straw that broke my back. They've just recently started looking at the upgrade. I can't even go through RPy to do calculations in R instead... because it's not available for 3.x yet either.
I really hope Python doesn't get hamstrung like Perl did. I switched to Python on purpose. Ruby, Scala, Haskell, Clojure, etc all have their pluses, but overall they'd be a step back for my work.
When I first heard about UnladenSwallow I cheered. Then read that it was not targeting Python 3, and realized that this is really throwing a wrench into the works for Python 3.
I mean, switching to Py3k for the clean-ups and features is nice, but it's going to be hard for many to migrate if they've got to leave a 5x improvement in speed behind.
The people behind UnladenSwallow certainly must realize the repercussions for Python 3, no?
Yes, the Unladen Swallow people know that Py3k is the future. They also know that Google has a shit ton of Py2.X code that it's their job to make fast. That's why they're developing against Python2.X and then when it's time to merge back into CPython it will be merged in trunk (2.X) and then forward ported into Py3k.
Let's think of Python 3 as Perl 6 that got done on schedule. There may be many good reasons to use Python 3 today, but it is not like Python 2.x is not good enough as-is or it is standing still.
With Guido's recent focus on not adding new semantics to the language, it's clear that he wants people to focus on getting stuff working on 3.x without changing more stuff in the core.
Optimizations to the language via unladden swallow, rewriting standard libararies etc., are a right step in that direction.
> Let's think of Python 3 as Perl 6 that got done on schedule.
That seems silly to me. Guido announced that he'd start working on Python 3000 part time in early spring 2000; that's eight and a half years.
Now compare the goals of Python 3 to the goals of Perl 6. Python 3 didn't really change much over Python 2, in comparison. Sure, Python 3 came out a little bit sooner than Perl 6, but it was also a lot less work to produce.
Every single library I've gone out to use wasn't compatible, and likely won't be this year. I've ported a couple, which wasn't hard, but silly that I had to do it. I've worked around the needs for others, but it's just not a reasonable way to work. Plus, when I've sought help, everyone balked at me using 3.
SciPy (NumPy) was the straw that broke my back. They've just recently started looking at the upgrade. I can't even go through RPy to do calculations in R instead... because it's not available for 3.x yet either.
I really hope Python doesn't get hamstrung like Perl did. I switched to Python on purpose. Ruby, Scala, Haskell, Clojure, etc all have their pluses, but overall they'd be a step back for my work.