What is there in 3 that would benefit scientific computing which 2.7 doesn't have? TBH I don't see the value proposition.

For web stuff you have asyncio, and for hackers there is the venv improvements, but what is there for scientific users who have work to get done?

And how many of them can't be backported to 2?

You don't need to freeze the language, but if the alternative is breaking backwards compatibility after all that time...

It's been seven years. Get over it already. That attitude keeps Python a fractured community.

This argument is just as valid as "You've been trying for 7 years. Give up already."