Meta-ticket: Review of Python 3 features that sagelib should use systematically

[mkoeppe]

For Sage 10.4, all of the following features can be used

Python 3.6 features (https://docs.python.org/3.10/whatsnew/3.6.html):

• pathlib
• static typing
  • add typing annotations in permutations #32058 add typing annotations in permutations
  • Developer manual: Section on typing annotations #32067 Developer manual: Section on typing annotations
  • Add a bit of typing to manifold code #29775 Add a bit of typing to manifold code
  • Python 3 typing/protocols/ABCs and categories #30381 Python 3 typing/protocols/ABCs
  • Type hints (__annotations__) and coercion/categories/Cython #30511 Type hints (__annotations__) and coercion/Cython
• native namespace packages - Modify find_python_sources to support modularization of sagelib by native namespace packages (PEP 420) #28925
• yield from - use yield from in poset code #29857
• f-strings (formatted strings)
• keyword-only parameters (Enforce keyword-only parameters, with deprecation #16607, sage_getargspec mishandles keyword-only arguments #31309)
• new import machinery - Make .sage modules importable #27074 (Make .sage modules importable)
• __matmul__ (infix @ operator) - Add support for __matmul__ in the coercion model #22760
• from ..a import b
• ...?

Python 3.7 features (https://docs.python.org/3.10/whatsnew/3.7.html)

• importlib.resources (sage.repl: Replace use of SAGE_EXTCODE by importlib.resources #31306)

• from future import annotations (Add more typing information to manifold package #31006)

• https://docs.python.org/3/library/dataclasses.html

Python 3.8 features (https://docs.python.org/3.10/whatsnew/3.8.html)

• assignment expressions := (https://docs.python.org/3.10/whatsnew/3.8.html#assignment-expressions)
• positional-only parameters (PEP 570)
• f'{expr=}' https://docs.python.org/3.10/whatsnew/3.8.html#f-strings-support-for-self-documenting-expressions-and-debugging
• PEP 590: Vectorcall: a fast calling protocol for CPython https://docs.python.org/3.10/whatsnew/3.8.html#pep-590-vectorcall-a-fast-calling-protocol-for-cpython
• functools.cached_property() https://docs.python.org/3.10/whatsnew/3.8.html#functools
• pprint(sort_dicts=False)
• shutil.copytree(dirs_exist_ok=True) https://docs.python.org/3.10/whatsnew/3.8.html#shutil
• as_integer_ratio, __index__ special method (https://docs.python.org/3.10/whatsnew/3.8.html#other-language-changes)
• math.dist, hypot, prod, perm, comb, isqrt https://docs.python.org/3.10/whatsnew/3.8.html#math
• typing.TypedDict, typing.Literal, Final, Protocol https://docs.python.org/3.10/whatsnew/3.8.html#typing

Python 3.9 features (https://docs.python.org/3.10/whatsnew/3.9.html)

Features introduced in Python 3.10 or later cannot be used in Sage yet because we support system Python 3.9.x

Python 3.10 features (https://docs.python.org/3.10/whatsnew/3.10.html):

• preparser + match-case = sadness #34678: match ... case
• avoid using itertools.pairwise #36609

Python 3.11 features (https://docs.python.org/3.11/whatsnew/3.11.html):

Python 3.12 features (https://docs.python.org/3.12/whatsnew/3.12.html):

Python 3.13 features (https://docs.python.org/3.13/whatsnew/3.13.html):

See also

• Meta-ticket: improve compatibility with the Python library and Python packages #29728: Meta-ticket: improve compatibility with the Python library
• Adopt the “time window-based” policy for support of Python versions from NEP 29 #30384: Adopt the “time window-based” policy for support of Python versions from NEP 29
• Drop Python 3.6 support #30551 Drop Python 3.6 support

---

CC: @fchapoton @dimpase @jhpalmieri @tobiasdiez @tscrim @nthiery @slel @nbruin

Component: python3

Keywords: sd109

Issue created by migration from https://trac.sagemath.org/ticket/29756

[egourgoulhon]

comment:2

About types, has a general policy been discussed on sage-devel?

[dimpase]

comment:3

typing is a standard Python 3 module, and it allows one to catch more errors.. Why would one have anything against it?

[kcrisman]

comment:4

typing is a standard Python 3 module, and it allows one to catch more errors.. Why would one have anything against it?

Are there reasons (e.g. performance) not to use it in certain circumstances?

[mkoeppe]

comment:5

I created this ticket as a place for discussions of these new features to make sure that we use them systematically in the Sage library, rather than in an ad-hoc way.

For example, for typing, I think there are opportunities to integrate it with the various dynamically generated classes that come in from the category system, as I noted in #29775.

[tobiasdiez]

comment:6

There shouldn't be any performance issues with typing as it is deactivated on runtime (the only additional overhead is the import of the typing package).
See also the discussion here https://realpython.com/python-type-checking/#pros-and-cons

[slel]

comment:11

Regarding f-strings, see this comment:

• #30107 comment:19

[mwageringel]

comment:18

Replying to @egourgoulhon:

About types, has a general policy been discussed on sage-devel?

+1 for using typing. Return types in particular are useful for tab-completion on the result of function calls. For example

sage: "hello".upper().lower().upper()

this tab-completes without evaluating the code (i.e. without setting IPython's IPCompleter.greedy config option).