Package: python-scipy Followup-For: Bug #723722 Apologies, I was trying to cancel my previous message but it slipped through by accident, so please ignore it.
What I found was that the warning is emitted in
all cases, whether
np.nanmean([np.nan, np.nan])
np.nanmean([[np.nan, np.nan]])
or np.nanmean(np.array([np.nan, np.nan]))
But it is only emitted the first time that np.nanmean is used.
