4 ms·

The author didn't really give any concrete examples. What kinds of major Python libraries still need to be ported over to Python3?

sitkack 12y ago
The author is talking about long tail libraries, not major ones. The thousands of libs on PyPi that are still on 2.x
- akerl_ 12y ago
  Except most of the more obscure libraries I find are either abandoned projects or were written to support the author's needs and don't fit the generic use case. To note, that's not a Python-specific issue: it's the nature of any language ecosystem where modules can be made by anyone, and I'm not even sure it's a bad thing: I've learned a lot, even from libraries that I didn't end up using for whatever reason. I think it's dangerous to imagine Python or any language as something where all the libraries you need always exist, and thus any time you have to write your own libraries it's a failure. Embrace the library ecosystem: you're going to eventually find something that isn't covered, which is when you write a library and participate. Disclaimer: I know open sourcing code isn't always possible, and that time/project constraints can impact how much backend code a developer can write. But if we dismiss Python 3 just because some of the more obscure libraries aren't ported, I wonder how we ever picked up Python 2 originally.
  - sitkack 12y ago
    I am not dismissing Python3 outright and neither is the OP. But rewriting millions of lines of code to satisfy an artificially imposed requirement is probably not a good idea even if the libraries are "abandoned" Python3 ranks high on the list of engineering failures. Python3 itself is fine, Python3 the deployment and migration is a disaster.
    - akerl_ 12y ago
      The abandoned libraries I find end up needed to be mostly rewritten, even if I were to use them on Python 2.
    - zo1 12y ago
      I think you over-estimate exactly how many lines of code it takes to port from 2 to 3. It's not that much, unless you plan on using all the new features of 3.

sitkack 12y ago

The author is talking about long tail libraries, not major ones. The thousands of libs on PyPi that are still on 2.x

akerl_ 12y ago
Except most of the more obscure libraries I find are either abandoned projects or were written to support the author's needs and don't fit the generic use case. To note, that's not a Python-specific issue: it's the nature of any language ecosystem where modules can be made by anyone, and I'm not even sure it's a bad thing: I've learned a lot, even from libraries that I didn't end up using for whatever reason. I think it's dangerous to imagine Python or any language as something where all the libraries you need always exist, and thus any time you have to write your own libraries it's a failure. Embrace the library ecosystem: you're going to eventually find something that isn't covered, which is when you write a library and participate. Disclaimer: I know open sourcing code isn't always possible, and that time/project constraints can impact how much backend code a developer can write. But if we dismiss Python 3 just because some of the more obscure libraries aren't ported, I wonder how we ever picked up Python 2 originally.
- sitkack 12y ago
  I am not dismissing Python3 outright and neither is the OP. But rewriting millions of lines of code to satisfy an artificially imposed requirement is probably not a good idea even if the libraries are "abandoned" Python3 ranks high on the list of engineering failures. Python3 itself is fine, Python3 the deployment and migration is a disaster.
  - akerl_ 12y ago
    The abandoned libraries I find end up needed to be mostly rewritten, even if I were to use them on Python 2.
  - zo1 12y ago
    I think you over-estimate exactly how many lines of code it takes to port from 2 to 3. It's not that much, unless you plan on using all the new features of 3.