Julia's Design Choices for Positional vs Keyword Arguments in Function Calls and Broadcasting

[haxfn]

Recently I have been feeling more and more ways to realise that julia is well designed. One instance of this is when I realized something about keyword arguments, and why non keyword arguments are unmentionable in function calls. I mean:

In python:

def myfun(a, *, b):
    print(a, b)

myfun(1, b=3) # does and should work
# myfun(1, 3) # does not and should not work

# but
myfun(a=1, b=3) # works but is confusing, in julia is illegal

Julia takes a clear stance on this, the reason for this to me, even still, was only partially motivated, until today when I found myself thinking about broadcasting over keyword arguments.

The docs mentioned this:

"Keyword arguments are not broadcasted over, but are simply passed through to each call of the function. For example, round.(x, digits=3) is equivalent to broadcast(x -> round(x, digits=3), x)."

I don't have a use-case for broadcasting over keyword arguments, but

(
    decision A=> keyword arguments unmentionable in function calls, 
    decision B=> broadcasting over keyword arguments is illegal
)

Could someone kindly clarify, I am just curious about both of the decisions and if there is any connection between both of these design choices?

[engineer-antonio]

Hi, amazing insight! The way Julia handles arguments is, in fact, one of the pillars that allows broadcasting and multiple dispatch to work so well together. In Python, the flexibility of calling positionals as keywords ends up generating unnecessary coupling with the internal names of variables. Julia sacrifices this 'freedom' in exchange for consistency and performance. Welcome to the team of those who appreciate the language's design!

[haxfn]

👏