Colour has been in public development for over 3 years. The package has grown in various directions since the initial release and offers a significant amount of features.

It is used in research groups such as the St Andrews HCI Research Group or companies like Google, Merck KGaA or The Moving Picture Company. Even though it has reached a certain stability and maturity, it is still in alpha development status.

Two important features are missing for a first feature complete stable release:

• The first one is that our current dictionary based spectral implementation has reached its limits when building support for Machado et al. (2010): attempting to alter the domain or range of a SpectralPowerDistribution is difficult. With that in mind, we have started to work on a new alternative implementation where data is exposed as a continuous function modeled using an interpolating function encapsulated within an extrapolating function: #335.

• The second is support for metadata inside the API. Most of the codebase adopts definitions/functions over classes to stay clean and lean, the aforementioned spectral implementation being a notable exception. As a consequence, it is hard to implement a non-intrusive classifying mechanism, provide usable hints on functions domain/range or create an auto-conversion layer. We have considered multiple ways of providing the necessary metadata, e.g., experimental/medatada* branches, and decided that the true elegant solution was through docstrings.

The following example showcases the current implementation, defining metadata for parameters, returns and the definition by using the notes section:

def luminance_Newhall1943(V):
    """
    Returns the *luminance* :math:`R_Y` of given *Munsell* value :math:`V`
    using *Newhall et al. (1943)* method.

    Parameters
    ----------
    V : numeric or array_like
        metadata : {'type': 'Munsell Value', 'symbol': 'V', 'extent': (0, 10)}
        *Munsell* value :math:`V`.

    Returns
    -------
    numeric or array_like
        metadata : {'type': 'Luminance', 'symbol': 'R_Y', 'extent': (0, 100)}
        *luminance* :math:`R_Y`.

    Notes
    -----
    metadata : {'classifier': 'Luminance Conversion Function', 'method_name':
        'Newhall 1943', 'method_strict_name': 'Newhall et al. (1943)'}

    References
    ----------
    .. [1]  Newhall, S. M., Nickerson, D., & Judd, D. B. (1943). Final report
            of the OSA subcommittee on the spacing of the munsell colors. JOSA,
            33(7), 385. doi:10.1364/JOSA.33.000385

    Examples
    --------
    >>> luminance_Newhall1943(3.74629715382)  # doctest: +ELLIPSIS
    10.4089874...
    """

    V = np.asarray(V)

    R_Y = (
        1.2219 * V
        - 0.23111 * (V * V)
        + 0.23951 * (V**3)
        - 0.021009 * (V**4)
        + 0.0008404 * (V**5)
    )

    return R_Y

There is one caveat though: running Python with -OO argument will optimize the bytecode and trim the docstrings and as a result preventing metadata usage. This is an edge case we are aware of and it will be advertised.

These two features are consequential and taking a fair amount of time to implement and test. They will also introduce backward incompatible changes.

Stay tuned!