Named Parameters vs. Positional Parameters

[willaaaaaa]

Take a side and stick with it to the bitter end.

[tetsujin]

Named can make calling interfaces very versatile and the significance of the parameters very easy to discern, and positional can make them very compact (which can also lead to improved clarity under certain circumstances).

So, both. Python is awesome.

(One case where I dealt with named parameters in Python was dealing with the serial library. I couldn't tell you the right order for the calling arguments when creating a serial port object, but supplying named arguments makes it pretty easy to get up and running without a major hassle.)

[ycc1988]

Named parameters allow for simple use of default values. For example, a plotting function may have default font faces, line styles, etc. MATLAB/Octave doesn't have true named parameters, but gets around this with some clever (i.e. annoying to implement) parsing.

[Xanthir]

Both. Positional parameters for the arguments that are absolutely necessary. Named optional params for *everything* else. And, for consistency, let the required parameters be nameable too.

I agree - Python is awesome in this respect.

[Quizatzhaderac]

If I had to pick one: positional.

That said I don't see good reasons not to have both. Though it does seem that positional has a relative advantage for strongly typed languages and named for weakly typed. In Java, a method that takes a bean, a string, an int, then a boolean is a lot clearer than a ruby method that takes an object, a second object, another object, then finally an object.

[Jplus]

I'm missing the otter/duck option. Named parameters can be useful but positional ones are fine too, so really I couldn't care less.