Awkward usage for layout_column_wrap

[wch]

The signature for layout_column_wrap is this:

def layout_column_wrap(
    width: Optional[CssUnit],
    *args: TagChild | TagAttrs,
    fixed_width: bool = False,
    heights_equal: Literal["all", "row"] = "all",
    fill: bool = True,
    fillable: bool = True,
    height: Optional[CssUnit] = None,
    height_mobile: Optional[CssUnit] = None,
    gap: Optional[CssUnit] = None,
    class_: Optional[str] = None,
    **kwargs: TagAttrValue,
) -> Tag:

This makes for kind of awkward usage. Because width comes before *args, it can never be named when the function is called (assuming you are passing in some *args, which you would always want to do). For example:

# These two don't work
layout_column_wrap(width="200px", "Hello", "world")
layout_column_wrap("Hello", "world", width="200px")

# This works
layout_column_wrap("200px", "Hello", "world")

On the other hand, height must always be named, and come after the *args. So if you want to specify both the width and height, you'd have to do this, which is awkward:

layout_column_wrap(1/2, "Hello", "world", height="300px")

Another awkward thing is that if you want to set width to None, you must call the function like this:

layout_column_wrap(None, "Hello", "world")

[gadenbuie]

Your timing on bringing this up is great. I was just looking at updating the CSS used by layout_column_wrap() in such a way that width would not be required. The idea would be to use an auto-fit grid layout set such that width defines the minimum space required to create a new column in a row and we otherwise allow grid to choose the number of rows and columns. (I created a placeholder PR with the core idea here, but won't get back to this until after conf.)

With this change, it would make a lot of sense to move width out of the first argument position on the R side too, which sounds like an improvement for both languages. My proposal for R is that width default to something like "200px", as an optional named argument, with all the same behavior if the user chooses to customize the argument.

This would be a breaking change on the R side, so now would be a great time for any additional feedback or collaboration around this function.