perf: Avoid redefining lambda in *Namespace.all

[dangotbanned]

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

• Mentioned in (chore: remove kwargs from selector function #2058 (comment))
• Originally (028098e) and (c597ba7)

Checklist

• Code follows style guide (ruff)
• Tests added
• Documented the changes

If you have comments or can explain your changes, please do so below

Previously, a new function was created in each of these locations - but it didn't bind any arguments - so there was no benefit to using a lambda

[MarcoGorelli]

thanks - could you clarify how this helps please? i don't understand the difference, and there aren't any arguments to bind in this function?

[dangotbanned]

thanks - could you clarify how this helps please? i don't understand the difference,

Sure thing @MarcoGorelli

Right now, every time all() is called that line is equivalent to:

def _(df):
    return df.columns
...
columns = _(df)

By equivalent - I mean if you call all() 3 times - that would expand to this:

def _1(df):
    return df.columns
...
columns = _1(df)
def _2(df):
    return df.columns
...
columns = _2(df)
def _3(df):
    return df.columns
...
columns = _3(df)

This change skips defining function each time, so those 3 calls would all be:

columns = get_column_names(df)
columns = get_column_names(df)
columns = get_column_names(df)

The difference is we didn't need to create a new inline function each time.

and there aren't any arguments to bind in this function?

What I mean is the scope created by lambda df: df.columns doesn't bind an argument in the definition.
By comparison, this lambda binds self._implementation, self._backend_version, self._version from the outer scope.

That would be a use-case I'd understand for a lambda - since it utilizes self

narwhals/narwhals/_pandas_like/namespace.py

Lines 125 to 132 in 8888e2c

	def all(self: Self) -> PandasLikeExpr:
	return PandasLikeExpr(
	lambda df: [
	PandasLikeSeries(
	df._native_frame[column_name],
	implementation=self._implementation,
	backend_version=self._backend_version,
	version=self._version,

Also some links:

• https://docs.python.org/3/reference/expressions.html#lambda
• https://docs.python.org/3/howto/functional.html#small-functions-and-the-lambda-expression
• https://discuss.python.org/t/what-is-the-purpose-of-lambda-expressions/12415/2

[MarcoGorelli]

sure, thanks @dangotbanned