decorated_options 0.1.0

Function decorator to make argument passing saner.

Function decorator to make argument passing saner.

I often have to write a function which runs a simulation/learning task which I need to run for several different parameters. This initially is manageable, but then slowly configuration creep starts to happen: I keep adding more and more parameters to the functions which run the simulations and keep making my old code more and more fragile.

I wrote decorated_options to decouple the arguments for different set of experiments.

In brief, decorated_options converts this:

def run(max_num_followers, num_segments, is_hawkes):
    # ...
    # ...

# tmp = run_multiple_followers(max_num_followers=10, num_segments=10, is_hawkes=True)
# tmp = run_multiple_followers(max_num_followers=100, num_segments=10, is_hawkes=False)
# tmp = run_multiple_followers(max_num_followers=10, num_segments=50, is_hawkes=True)
tmp = run_multiple_followers(max_num_followers=1000, num_segments=100, is_hawkes=False)

to:

from decorated_options import Options, optioned

@optioned('opts')
def run(max_num_followers, num_segments, is_hawkes):
    # ...
    # ...


opts = Options(max_num_followers=10, num_segments=10, is_hawkes=True)
# tmp = run_multiple_followers(opts=opts)
# tmp = run_multiple_followers(max_num_followers=100, is_hawkes=False, opts=opts)
# tmp = run_multiple_followers(num_segments=50, is_hawkes=False, opts=opts)
tmp = run_multiple_followers(max_num_followers=1000, num_segments=100, is_hawkes=False)

• Benefits over **kwargs in receiving function:

  1. Early reporting of errors at call-time.

  2. No need to unpack the values.

  3. Default values do not have to be hard-coded.

  4. Allows progressive improvement, no need to change old code which uses positional arguments.

• Benefits over **dict while calling:

  1. Easier updating/overriding of values

  2. Positional arguments also work

  3. Guaranteed immutability (throws Exceptions on attempted violations.)

• Benefits over default values in receiving function:

  1. Options objects can save defaults for multiple settings.

  2. De-couples default values from the functions themselves.

Installation

pip install git+https://github.com/musically-ut/decorated_options.git@master#egg=decorated_options

Development

To run the all tests run:

tox

Note, to combine the coverage data from all the tox environments run:

Windows	set PYTEST_ADDOPTS=--cov-append tox
Other	PYTEST_ADDOPTS=--cov-append tox

Changelog

0.1.0 (2016-07-24)

• First release on PyPI.