[ghshephard]

It's challenging. On the one hand, you don't want to break what is working (Python 2.x), so you'd like to support it so that community can continue to be productive. On the other hand, you don't want your braintrust split across two different versions, and you don't want the confusion that comes from having two major versions.

For example, the (Epic) Learning Python 5th Edition by Mark Lutz, had to be written for two simultaneous audiences, 2.7 and 3.3. There are all these disclaimers noting where things are different between the two versions, and there is a lot of cognitive overload trying to read a book can't assume you are on Python3.

In the best case world - people would treat this like a y2k situation, and realize that if they didn't get with the program, and migrate over to Python3, they'll end up like Perl, with some other newcomer that isn't so bipolar charging forward and winning mindshare.

Unfortunately, their are a lot of Python2 people who are happy with Python2, and we're in the situation we have today.

[melling]

There are a lot of Windows users who are happy with XP too. I remember trying to explain to people 20 years ago why HDTV was better than what they had. Most people were happy with 480 lines of resolution, or thought they were. It's human nature. By being accommodating, you've made the problem harder.

Apple is the only company that says screw legacy. Of course users complain but they just grumble and know to accept it. Apple and their users, as a whole, benefit greatly by "getting all the wood behind one arrow" strategy.

There's lots of grumbling about Swift but I bet we get a very large developer adoption rate within 24 months.

[ghshephard]

But Swift is an actual example of something with no legacy to worry about. Instead of being Objective C++2015, it's an entirely new language. And, more importantly, Apple can give guidance that people have to follow. As Swift matures, regresses, and has the bugs worked out of it, and library support is fully developed, Apple can arbitrarily identify a point at which Objective C just has to die, and the developers (if they want to develop applications for the App Store, which is 99% of them), will just have to do so.

The nice thing about Windows XP, is that by and large, a Windows XP user could use Windows 7 with next to zero training. Or, in fact, zero training.

This is why the first version of Windows 8 pissed so many people off - removing the Start Button all of a sudden made the Operating System unusable to a huge portion of the Windows user base.

I understand they've brought back the start button.

[qznc]

Swift has legacy builtin. It is forced to use Objective-C's OO mechanisms.