The API of the new PhotoCalib object is entirely different from the old Calib API: it is defined in terms of a flux calibration factor instead of a magnitude zero point, methods are named e.g. instFluxToMagnitude instead of getMagnitude, and it never raises an exception when given a negative instrumental flux. To ease the transition to PhotoCalib, I propose that PhotoCalib gain the following deprecated methods from Calib until v18.0:

    /**

     * Return an instrumental flux and error (in ADU) given magnitude and magnitude error.

     *

     * Assumes that the errors are small and uncorrelated.

     *

     * @param mag The AB magnitude of the object.

     * @param magErr The error in the magnitude.

     */

    std::pair<double, double> getFlux(double const mag, double const magErr) const;

    ndarray::Array<double, 1> getFlux(ndarray::Array<double const, 1> const& mag) const;

    std::pair<ndarray::Array<double, 1>, ndarray::Array<double, 1>> getFlux(

            ndarray::Array<double const, 1> const& mag, ndarray::Array<double const, 1> const& magErr) const;

    /**

     * Return a magnitude and magnitude error given instrumental flux and error (in ADU).

     *

     * @param instFlux The measured instFlux of the object (ADU).

     * @param instFluxErr The error in the measured instFlux (ADU).

     */

    std::pair<double, double> getMagnitude(double const instFlux, double const instFluxErr) const;

    ndarray::Array<double, 1> getMagnitude(ndarray::Array<double const, 1> const& instFlux) const;

    std::pair<ndarray::Array<double, 1>, ndarray::Array<double, 1>> getMagnitude(

            ndarray::Array<double const, 1> const& instFlux,

            ndarray::Array<double const, 1> const& instFluxErr) const;

Other methods on Calib that we will not provide backwards-compatible deprecated methods for:

• setFluxMag0: Invalid; PhotoCalib is immutable.
• getFluxMag0: PhotoCalib has getInstFluxAtZeroMagnitude, which only returns the 0-point and not the 0-point error, which has a non-trivial relationship with the calibration error. This was most commonly used to construct a new Calib, about which see the next point.
• Construction from a "instFlux at zero magnitude" double: this cannot be distinguished from the existing calibrationMean constructor. If such functionality is strongly desired, we could add a makePhotoCalibFromZeroPoint (or similar) free function, but I doubt it is necessary: users rarely have to construct their own photoCalib, and PhotoCalib will be able to read older Exposure files that contain persisted Calibs.
• get/setThrowOnNegativeFlux: Invalid; PhotoCalib is immutable and never throws in this manner.
• operator*= and operator/=: Invalid; PhotoCalib is immutable. We could provide operator* and operator/ though (returning a new PhotoCalib), if this functionality was important.