Committing a piecewise exponential interactive fit raises ValueError: Unsupported fit type

[PierreRaybaut]

Summary

commit_interactive_fit() derives the Sigima fit type from the menu id:

dst.metadata["fit_params"] = _fit.create_fit_params(
    fit_id.removesuffix("_fit"), ...
)

For "piecewiseexponential_fit" this yields "piecewiseexponential", which is not a key of Sigima's FIT_TYPE_MAPPING. The 14 valid keys are:

cdf, doubleexponential, exponential, gaussian, linear, lorentzian,
multigaussian, multilorentzian, planckian, polynomial, sigmoid,
sinusoidal, twohalfgaussian, voigt

The menu id was chosen to match DataLab desktop's piecewiseexponential_fit, but the underlying computer is DoubleExponentialFitComputer, whose Sigima type is doubleexponential. Committing the fit therefore fails with ValueError: Unsupported fit type.

Location

src/runtime/dlw_interactive_fit.py — _INTERACTIVE_FITS entry and commit_interactive_fit().

Proposed fix

Store the Sigima fit type explicitly on _FitKind instead of deriving it from the menu id by string surgery, and default it to fit_id.removesuffix("_fit") for the other twelve entries.

Root cause

tests/python/test_interactive_fit.py only exercises commit_interactive_fit() for gaussian_fit. A test parametrised over _INTERACTIVE_FITS would have caught this. The fix must add that loop.