Require Keyword Arguments

July 13th, 2014
python, tech
To avoid errors in calling a python function you might want to ensure it's only ever called with keyword arguments. Say you have:
    def score(actual, predicted):
       ...
and you're worried people accidentally call it with the arguments reversed. After all, nothing looks surprising with:
    score(model.predict(data), isRepaid)
We can have a culture where everyone writes in the keyword arguments, as:
    score(actual=isRepaid, predicted=model.predict(data))
This mostly solves the problem, but could we have Python check this for us? Yes! PEP 3102 added this to Python 3+:
    def score(*, actual, predicted):
      ..
But what about the Python 2 series? You could rebuild Python with this patch applied, or you could use a decorator that verifies there are no positional arguments and then calls the original function:
    def poscheck(f):
      def checked_f(*args, **kwargs):
        if args:
          raise PositionalArgumentsError(f)
        f(**kwargs)
      return checked_f

    @poscheck
    def score(actual, predicted):
      ...
This is pretty good, but what about a case like this:
    # Split s on newlines ('\n'), returning a list.  These newlines
    # are not normally retained, but if the optional keepends argument
    # is True then they're kept.
    def splitlines(s, keepends=False):
       ...
If someone calls splitlines as splitlines(s, True) that's going to be pretty confusing to a reader. You'd like to make sure writers always make it clear what the boolean is about, so you have a house style where you write splitlines(s, keepends=True). If you used poscheck, though, that would require splitlines(s=s, keepends=True) which is too verbose. So, use poscheck_except instead:
    @poscheck_except(1)
    def splitlines(s, keepends=False):
      ...
This will require that after the first positional argument all other arguments are given with keywords, if present.

The code is poscheck on github.

Comment via: google plus, facebook, substack

Recent posts on blogs I like:

It isn’t always evil to ally with Steve Bannon

Scott Alexander wrote:

via Thing of Things October 19, 2026

Was the HuggingFace hacking incident scheming?

In July 2026, internal OpenAI agents hacked HuggingFace (HF) in order to figure out how their evaluation grader works. This involved a thousand agents coordinating on an improvised message board to carry out an elaborate sequence of exploits. A later grou…

via Victoria Krakovna October 18, 2026

There's no point at which turning your brain off will work

In early 2025, I started seeing people turn off their brain as they use LLMs1. They would have an LLM take an action (summarize text, write some code, etc.), and just assume that it worked2. This generally didn't work in early 2025 and the result was …

via Posts on September 18, 2026

more     (via openring)