Add min_weight param to rolling_exp functions

[max-sixty]

No description provided.

[Illviljan]

Is this a lazy import? Does our tests catch it?

Could you add numbagg here so we don't regress with regards to #6726:

https://github.com/pydata/xarray/blob/e8be4bbb961f58ba733852c998f2863f3ff644b1/xarray/tests/test_plugins.py#L216C9-L239

You can also use this for checking module availability:

xarray/xarray/namedarray/utils.py

Lines 81 to 84 in e8be4bb

	def module_available(module: str) -> bool:
	"""Checks whether a module is installed without importing it.
	Use this for a lightweight check and lazy imports.

[max-sixty]

It's not a lazy import per se (it used to be, but I'm not sure that was necessarily deliberate vs. coincidental)

Using module_available doesn't seem to get the version?

What's the standard way of getting an optional module and checking the version in xarray? I'm happy to abide by the standard if there is one. The current approach does seem to work well in isolation...

[Illviljan]

You can use

xarray/xarray/core/pycompat.py

Lines 81 to 83 in e8be4bb

	def mod_version(mod: ModType) -> Version:
	"""Quick wrapper to get the version of the module."""
	return _get_cached_duck_array_module(mod).version

To get the version.

[max-sixty]

Though this isn't a duck array module — numba / numbagg operates on numpy arrays. Would I still use this?

[max-sixty]

The tests fail because the build is cached on an old version of numbagg, we can try again when that's updated

[max-sixty]

Let me know any feedback re the version handling. Otherwise ready to go, will merge in the next day or so (and then happy to make any further changes from there if we don't catch something beforehand) — hope that's reasonable