This article is a little light on substance. Aside from the now mandatory brackets around print() you can write Python 3 code that is indistinguishable from Python 2 code. Sure, Python has become more complex and the standard library has grown substantially, but using Python for one-off scripts is as easy as it has ever been.
If you don't want to deal with async stuff... then just don't use it. In the last week or two, I started a new personal project in Python, one where I decided from the outset to take full advantage of 3.6, and I don't have a single async, await, import asyncio, or @asyncio.coroutine anywhere in my code. It's not the kind of code I want to write, so I don't write it. Instead, I'm happy with f-strings and the ability to union dicts together [0].
I do agree though that the type annotations hurt the overall austere laconicity of Python code. I don't use them myself, and I'm lucky that I don't have to read through anyone else's code that uses them, so it's not quite a big deal (yet).
f-strings alone have made me not want to go back to any older version.
[0] {**foo, **bar} # where foo and bar are both dicts; introduced in Python 3.5
I use a lot of software written by other people. I need to, in many cases, read and understand that software. So pretending it doesn't exist is fine for personal, small projects, but doesn't seem as realistic in my day-to-day work.
Also, this post wasn't about every little feature in Python 3. It was about 2 features that, I feel, make python as a language less pythonic.
To those that downvoted, care to explain? I think it's a reasonable point to make that many of us read as much (or more) code as we write. Therefore, my concerns about readability.
The article presents a blatant strawman argument against Python 3's language by using advanced language features to claim it's no longer simple nor suitable for beginners.
Just don't use the advanced features if you don't need them, the core language is the same as Python 2 which the author lauds as being wonderful.
Including, incidentally, in the (invalid by its own description) example, function decorators and the format method on strings, which were both present in Python 2.
This is similar to how I feel. Sometimes I tell people I'm a Python 1.5 guy, because that's the version I started with. I've since familiarized myself with list comprehensions, generators, etc. I even use decorators, though I'm still at the point where I have to find a howto and refresh some of the details each time. The print and string changes I find annoying, and I haven't gotten to async/await yet. It's not that I don't understand or appreciate these things, but they do take away a lot from Python's simplicity and intuitive readability. Those were massive strengths, so I'm not sure the tradeoff is worth it. There are already plenty of languages for those who prefer code which some would call elegant and others (including me) would call obtuse. I find myself looking more and more at Lua, which seems to embody more of the early Python spirit than modern Python itself does.
I think that I'll have to agree mostly with article.
At first I thought async/await would be a good feature, but after studying it a bit more and trying to use it in a real project, I've become very disappointed.
There was really no need to add new keywords to the language. It could have been more transparent like generators (it's a generator if it has a yield).
And, the implementation of coroutines as a syntactic sugar on top of generators is like buying an SUV and finding out later that it's a minivan chassis with bigger wheels.
11 comments
[ 4.1 ms ] story [ 55.0 ms ] threadI do agree though that the type annotations hurt the overall austere laconicity of Python code. I don't use them myself, and I'm lucky that I don't have to read through anyone else's code that uses them, so it's not quite a big deal (yet).
f-strings alone have made me not want to go back to any older version.
Also, this post wasn't about every little feature in Python 3. It was about 2 features that, I feel, make python as a language less pythonic.
Just don't use the advanced features if you don't need them, the core language is the same as Python 2 which the author lauds as being wonderful.
At first I thought async/await would be a good feature, but after studying it a bit more and trying to use it in a real project, I've become very disappointed. There was really no need to add new keywords to the language. It could have been more transparent like generators (it's a generator if it has a yield).
And, the implementation of coroutines as a syntactic sugar on top of generators is like buying an SUV and finding out later that it's a minivan chassis with bigger wheels.