Closed jgieseler closed 1 year ago
(Hopefully) fixed with https://github.com/jgieseler/solo-epd-loader/commit/7e200042337e4d1acd8b95a809b12a4744f92374
Pandas seems to became really picky and needs now np.float32(-1e+31)
instead of just -1e+31
(which will be just float
) in the replace
function.
NB: Code changed with using sunpy 4.1.0 because that replaces FILLVALs directly, so no need anymore to do it manually here.
At least until pandas 1.4.4 the replacement of FILLVAUES done by the following code worked: https://github.com/jgieseler/solo-epd-loader/blob/f92e4e995a273d5755792c3f02e4ea3c33cfc675/solo_epd_loader/__init__.py#L754-L761
But since pandas 1.5.0 it doesn't work anymore, and the values of -1e+31 are not replaced with np.nan's.
I don't know the reason, maybe it has to do with the fact that the corresponding DataFrames have a MultiIndex.