What should be in the Python standard library?

Posted Jan 10, 2019 14:58 UTC (Thu) by mageta (subscriber, #89696)
Parent article: What should be in the Python standard library?

I'd not put it past them to now that people just got over the pain of the python 2->3 migration go and break tons of scripts again because they minimize the stdlib. Have they really learned nothing?

What should be in the Python standard library?

Posted Jan 10, 2019 17:38 UTC (Thu) by smurf (subscriber, #17840) [Link]

Presumably, if you get your Python from a distribution, the recommended set will include the pieces that have been removed from stdlib.

Also, there are downsides to having a large stdlib. Presumably² the Python developers are able to reason about them and strike a, well, reasonable balance between maintainance burden and compatibility issues, esp. since this would fall flat on its face at script startup time instead of crashing some indeterminate time later.

What should be in the Python standard library?

Posted Jan 10, 2019 23:00 UTC (Thu) by NYKevin (subscriber, #129325) [Link] (1 responses)

I disagree. I think it's pretty clear that Python 3 was an exceptional situation, and that going forward, they intend to adhere to PEP 4 for future deprecations. Furthermore, I'm rather skeptical that a significant number of actually used modules are going to be removed any time soon, even if they are deprecated. (For example, macpath is slated for removal in 3.8, but I doubt anyone cares at this point. On the other hand, there is no indication that they intend to remove optparse any time soon.)

However, I should also point out that the table of contents is getting unwieldy. It might make sense to reorganize it, or to split it into multiple pages. That would not break anyone's old scripts.

What should be in the Python standard library?

Posted Jan 10, 2019 23:26 UTC (Thu) by karkhaz (subscriber, #99844) [Link]

> It might make sense to reorganize it, or to split it into multiple pages. That would not break anyone's old scripts.

It breaks the flow of navigating to that page and browser-searching for a word related to the module that you want to use, whose name you do not know. For me, this has always been a reliable way of finding the right module.

If you split the ToC into multiple pages, I now need to guess what arbitrary page or category somebody has placed a module into, browse to that, search, discover that my guess was wrong, navigate back up and try again, and by now my concentration is long gone.

The ToC is not broken, there is no need to fix it.