The next 20 years of Python [LWN.net]

The next 20 years of Python

Posted Jul 11, 2013 6:39 UTC (Thu) by vedantk (guest, #88435)
Parent article: The next 20 years of Python

> Cutting corners is allowed, as you can always go back and improve it later.

Some of the aspects of Python that I've grown to love the most were introduced later on during its lifetime. Take the collections module for example (2.4), or the nifty "yield from" statement (3.3).

My impression is that the language is ambitious and improving. It should fare well during the foreseeable future.

to post comments

The next 20 years of Python

Posted Jul 11, 2013 9:43 UTC (Thu) by renox (guest, #23785) [Link] (5 responses)

>My impression is that the language is ambitious and improving. It should fare well during the foreseeable future.

Improving yes, but very slowly so I'm not that sure:
-languages with optional static typing seems to have "the best of both world" that Python has not.
-performance: performance of Python on one processor isn't that great.
-performance bis: using several CPU in Python isn't that great either.

That said, it is very,very hard for new languages to gather momentum so this should ensure that Python is used for a long time, but there's a difference between being very used but dying like Cobol/Perl(not a troll: that's *my* opinion YMMV) and being very used and growing..

The next 20 years of Python

Posted Jul 11, 2013 12:15 UTC (Thu) by smitty_one_each (subscriber, #28989) [Link] (3 responses)

>languages with optional static typing

I'm curious how your comment differs from functionality available via
http://wiki.python.org/moin/PythonDecoratorLibrary#Type_E...

The next 20 years of Python

Posted Jul 11, 2013 12:26 UTC (Thu) by mpr22 (subscriber, #60784) [Link] (2 responses)

Static typing isn't just about detecting passing an int to something that expects a float (or vice versa), it's about detecting passing a Penguin to something that expects a Duck. I concede that it may be the case that the decorator you linked to supports the latter check and just has laughably simplistic examples.

The next 20 years of Python

Posted Jul 11, 2013 14:09 UTC (Thu) by juliank (guest, #45896) [Link] (1 responses)

Well, it does, but it's still only a run-time check, and those are not very useful. And in Python 3, you can write things like:

def f(x: int, y: int, z: int) -> int:
return x * y * z

Combined with a decorator or a meta-class, you can use this for run-time type checking as well. But it's still only at run-time, and not static, at compile-time.

The next 20 years of Python

Posted Jul 22, 2013 6:28 UTC (Mon) by smurf (subscriber, #17840) [Link]

Well, Python isn't compiled, strictly speaking, so that's hardly a surprise.

However, once you do have those annotations, running an analyzer over the bytecode which verifies a large subset of them shouldn't be *that* hard to write.

The next 20 years of Python

Posted Jul 17, 2013 5:37 UTC (Wed) by Cato (guest, #7643) [Link]

Python seems to be gaining momentum not so much in the core language as the libraries and application areas - some very large web apps such as Instagram and Disqus (25,000 requests/sec) are built in Python.

Python is also gaining a lot of use in scientific computing where the ability to rapidly explore solutions is a big win and there are various compilers and JITs (Numba, Cython, PyPy). You can even run Python on CUDA based GPUs if your problem fits.

There is still the issue of migrating all major libraries to Python 3, but there's now a lot more experience and tooling for this.

Links:

- Disqus: world's largest Django app at 400m users - http://ontwik.com/python/pycon-2011-disqus-serving-400-mi...
- PyCUDA: http://mathema.tician.de/software/pycuda