Unladen Swallow: Python 3’S Best Feature(jessenoller.com)
jessenoller.com
Unladen Swallow: Python 3’S Best Feature
http://jessenoller.com/2010/01/06/unladen-swallow-python-3s-best-feature/
24 comments
http://twitter.com/gvanrossum/status/7464557748 BDFL likes
I imagine BDFL would do anything at this point to get Py3k jumpstarted.
That assumes any of us (BDFL included) are desperate to get it in production Right Now, which isn't true. However, backwards incompatibility is a tough pill to swallow and while Py3k is a cleaner, saner language it has still been a tough sell.
This should give it a helpful kick in the pants.
This should give it a helpful kick in the pants.
As far as I'm concerned (as a CPython developer), the 2.x branch can live as long as necessary. There is no reason to play games trying to push people to upgrade to 3.x. Those tactics piss people off and for good reason.
That said, I don't see this as a tactic to get people to upgrade. Python is an open source project and the developers get to choose what they will work on. If the Unladen Swallow guys want to merge into 3.x, that's their right. If someone else steps up to merge it into 2.x and support it, that's their right too.
Myself, I'm going to continue to fix bugs in the 2.x stream for the foreseeable future. I do plan on moving my applications to 3.x eventually though.
That said, I don't see this as a tactic to get people to upgrade. Python is an open source project and the developers get to choose what they will work on. If the Unladen Swallow guys want to merge into 3.x, that's their right. If someone else steps up to merge it into 2.x and support it, that's their right too.
Myself, I'm going to continue to fix bugs in the 2.x stream for the foreseeable future. I do plan on moving my applications to 3.x eventually though.
As a Cpython developer; you and I both know 2.7 is slated to be the final "big" release of python 2.x. Anything past that point will be only bug fix/security fixes.
Unladen would require a whole new 2.x release, which is not, and more than likely will not be planned. We're not playing games, we're putting features into 3k knowing full well we plan on putting 2.x out to pasture. We're making the selling points of Py3k porting much more compelling to library/framework/etc authors.
If someone wants to fork 2.7 and merge it with unladen, or better yet simply run the 2.x release of unladen-swallow up on google code, that's totally their prerogative, and I know some people will. From a feature standpoint though, this is a good think for 3k as it sits today.
Unladen would require a whole new 2.x release, which is not, and more than likely will not be planned. We're not playing games, we're putting features into 3k knowing full well we plan on putting 2.x out to pasture. We're making the selling points of Py3k porting much more compelling to library/framework/etc authors.
If someone wants to fork 2.7 and merge it with unladen, or better yet simply run the 2.x release of unladen-swallow up on google code, that's totally their prerogative, and I know some people will. From a feature standpoint though, this is a good think for 3k as it sits today.
I honestly thought py3k wasnt being sold. As of the last pycon, there was a lot of talk of 'slow and steady py3k adoption'. The general roadmap assumed by most people I interact with is: at some point someone will make a killer feature with py3k and that will be the tipping point. Until then, 2.x doesn't stop being awesome just cuz 3 around.
There has always been discussion about adding features to 3k to increase adoption speed. Even when unladen goes into 3k, which is probably the middle-to-end of this year, it's another year before major OS vendors pick it up, etc.
Nothing happens overnight, this is just something to incentivise possible porting efforts.
Nothing happens overnight, this is just something to incentivise possible porting efforts.
What is the killer app for python though? Django? numpy?
I don't think it really has one, certainly not like the ruby/rails relationship.
Python might not have a single "killer-app", but it has a killer ecosystem, which I learned to value higher.
Much of the joy in working with python comes from the "already written" effect. Libraries exist for almost any imaginable task, and it doesn't cease to surprise me how good the average code-quality is. It's extremely rare for me to run into a foul egg, which is a pleasant change from the platforms I've previously worked with.
Much of the joy in working with python comes from the "already written" effect. Libraries exist for almost any imaginable task, and it doesn't cease to surprise me how good the average code-quality is. It's extremely rare for me to run into a foul egg, which is a pleasant change from the platforms I've previously worked with.
I agree and I love python for that, but people learn Ruby specifically to do Rails work.
My point is only that python has no similar library or framework driving uptake.
My point is only that python has no similar library or framework driving uptake.
So far it hasn't been hurt by that. Every time I read about scripting language popularity python is usually right up there, in front of all others.
Only because the python community is not as vocal as, say, the rails community doesn't mean it has no uptake.
Only because the python community is not as vocal as, say, the rails community doesn't mean it has no uptake.
Don't forget the Natural Language Toolkit ( http://www.nltk.org/ ).
Rails is just a web framework. Killer? Yes. All you need for many apps? No.
And yes, both Django and numpy are also killer packages ... with Django probably being the inspiration for merging Rails with Merb.
Rails is just a web framework. Killer? Yes. All you need for many apps? No.
And yes, both Django and numpy are also killer packages ... with Django probably being the inspiration for merging Rails with Merb.
Up until now, it didn't seem like there were that many compelling reasons to try to jump to Python 3 - most of the changes were for correctness. Something like this gives a tangible reason to at least consider the pain of a major upgrade.
What does SGTM stand for?
What does SGTM stand for?
Sounds good to me.
Sounds good to me.
In firefox, you can just type "slang sgtm" and you get to http://www.urbandictionary.com/define.php?term=sgtm
It's the definitive quickreference on slang.
It's the definitive quickreference on slang.
I thought Unladen Swallow had good reasons to be on 2.x, such as the fact that Google is on 2.x internally. Maybe Google's decided it's time to migrate?
EDIT: the FAQ I was thinking of is at the top of http://code.google.com/p/unladen-swallow/wiki/FAQ . They do say 2.x internal, but willing to port to 3.x to get it merged with mainline.
EDIT: the FAQ I was thinking of is at the top of http://code.google.com/p/unladen-swallow/wiki/FAQ . They do say 2.x internal, but willing to port to 3.x to get it merged with mainline.
This is pretty amazing news. Last I heard, I thought that UnladenSwallow was 2.x-only. Now it turns out that it's going into 3.x and there will be no 2.x Unladen. Wow.
Really great news for Py3k.
Really great news for Py3k.
Does this mean Unladen Swallow is giving up? Just kind of liquidating the useful things they made into Python proper?
Giving up! Quite the opposite, the goal the entire type was for Unladen Swallow to be merged back into Python itself, so everyone can benefit. Bringing it back into CPython gives it the potential for the most eyeballs. I don't know whether Google will continue to pay devs to work on it full time, but certainly the community won't let it languish!