4 ms·

I don't think Python packaging is as bad as people make out (if packages only stick to the basic features of Python!) A far bigger issue I see is despite Python

I don't think Python packaging is as bad as people make out (if packages only stick to the basic features of Python!) A far bigger issue I see is despite Python supposedly being 'cross-platform': the language introduces so many small, backwards-compatibility breaking changes that you really need to use the most up to date interpreter you can. For instance: there is now a really cool operator that lets you do expressions where you assign a value and evaluate the result in one line. So you can write the C equivalent of something like while(buf = recv(...)) ... and it will work. Well, any program that uses this new feature won't run on anything but the most bleeding edge Python versions. Despite a program only needing a small fix to remove such expressions. It wouldn't run at all. I think the Python interpreter needs to be 'smarter' and add the capability to automatically install parallel Python versions if it detects a package using a more recent interpreter. Would honestly solve so many issues.

rtev 4y ago
While I do see the value in something like that, I can’t help but feel like that would just bake in more complexity. The python ecosystem is fragmented with micro fixes and half-solutions without anything truly addressing the major confusing aspects.
bborud 4y ago
There's what people think and then there is observed reality. Observed reality informs us that this is a big problem with Python, so we'll just have to accept that. What people could do isn't the issue - what people do is. Also, you don't fix a problem by adding complexity and unpredictability. Again, as reality informs us, this hasn't worked for Python. It is still a mess.
diarrhea 4y ago
Python 3.8 and its walrus operator is 2.5 years old, it's not bleeding edge, at least not anymore. That being said, I'm worried about the language in that complexity increases through the introduction of new language features without much benefit at all.
- morelisp 4y ago
  2.5 years is still bleeding edge for a backwards-incompatible syntax change.
  - stavros 4y ago
    It's not a syntax change, it's a new feature. If you know your version of Python doesn't support it, don't use it. I think your issue is mostly with library authors who drop older versions just so they can use a slightly more convenient syntax.

rtev 4y ago

While I do see the value in something like that, I can’t help but feel like that would just bake in more complexity. The python ecosystem is fragmented with micro fixes and half-solutions without anything truly addressing the major confusing aspects.

bborud 4y ago

There's what people think and then there is observed reality. Observed reality informs us that this is a big problem with Python, so we'll just have to accept that. What people could do isn't the issue - what people do is. Also, you don't fix a problem by adding complexity and unpredictability. Again, as reality informs us, this hasn't worked for Python. It is still a mess.

diarrhea 4y ago

Python 3.8 and its walrus operator is 2.5 years old, it's not bleeding edge, at least not anymore. That being said, I'm worried about the language in that complexity increases through the introduction of new language features without much benefit at all.

morelisp 4y ago
2.5 years is still bleeding edge for a backwards-incompatible syntax change.
- stavros 4y ago
  It's not a syntax change, it's a new feature. If you know your version of Python doesn't support it, don't use it. I think your issue is mostly with library authors who drop older versions just so they can use a slightly more convenient syntax.