First example in docs doesn't work

[eddy-oj]

from ncplot import view
view("https://psl.noaa.gov/thredds/dodsC/Datasets/COBE2/sst.mon.ltm.1981-2010.nc", coast = True)

Error:

File /usr/local/lib/python3.10/dist-packages/holoviews/core/sheetcoords.py:518, in Slice._boundsspec2slicespec(boundsspec, scs)
515 t_m,l_m = scs.sheet2matrix(l,t)
516 b_m,r_m = scs.sheet2matrix(r,b)
--> 518 l_idx = int(np.ceil(l_m-0.5))
519 t_idx = int(np.ceil(t_m-0.5))
520 r_idx = int(np.floor(r_m+0.5))

ValueError: cannot convert float NaN to integer

[robertjwilson]

Thanks. It looks like an update to xarray has broken this example. I will rewrite the internals to get it working. In the meantime, there is a new version of the example which should work if you have up to date packages. The only thing you need to change is to specify the variable you want to plot.

https://ncplot.readthedocs.io/en/latest/howto.html

Internally, the issue is due to how ncplot identifies which variables to plot. Previously, it was only identifying sst and plotting it, but now it's finding more variables. For some reason xarray lists 3 data variables for this dataset, when to me it should only have sst. This issue does not affect plotting in nctoolkit, which is the reason for ncplot existing as a separate package. Internally, nctoolkit uses CDO to identify variables to plot, and it only lists sst.

I will leave this open until I modify the code to identify the variables correctly.