switches MAD calculations from scipy to astropy, thus running away from a weird error where scipy changed the name of the function between versions that users might reasonably have installed in their environments. This closes #185
adds a warning for folks trying to bin an unnormalized Rainbow object. In some cases it might be fine to do so, but if there are uncertainties that vary significantly between wavelength bins, the current bin function will do well only if fluxes are normalized to be close to 1. We should revisit this question later, but for now at least be transparent. This closes #184 , and addresses a problem that @Pat-Wachiraphan ran into with some NIRISS data.
fixes, hides, and/or explicitly checks most of the expected warnings in the tests. Previously, there were a lot of noisy warnings about things that didn't matter. Having gotten rid of them, there's just one xarray deprecation warning that should still be dealt with at some point soon
changes the formatting for chromatic-based warnings to 🌈 suggestion (since that's what most of them are) and makes this custom format temporary, so it no longer permanently changes the appearance of warnings coming from other packages or any time after chromatic has been imported. This closes #186
adds h5py to the dependencies, based on @will-waalkes 's recent re-install experience
This pull request:
scipy
toastropy
, thus running away from a weird error wherescipy
changed the name of the function between versions that users might reasonably have installed in their environments. This closes #185Rainbow
object. In some cases it might be fine to do so, but if there are uncertainties that vary significantly between wavelength bins, the currentbin
function will do well only if fluxes are normalized to be close to 1. We should revisit this question later, but for now at least be transparent. This closes #184 , and addresses a problem that @Pat-Wachiraphan ran into with some NIRISS data.xarray
deprecation warning that should still be dealt with at some point soonchromatic
-based warnings to🌈 suggestion
(since that's what most of them are) and makes this custom format temporary, so it no longer permanently changes the appearance of warnings coming from other packages or any time afterchromatic
has been imported. This closes #186h5py
to the dependencies, based on @will-waalkes 's recent re-install experience