mfouesneau / ezdata Goto Github PK

Would be good to know where we are in this process.

Tick labels are not shown when plotting a corner or pairgrid.

It seems a Colab issue, but cannot find what this is. I cannot reproduce the issue when running in Jupyter notebook.

plt.setp(get_ticklabels(), visible=True) does not work. It actually seems like the Text values are removed.

If one reverses the axis, the result shown is empty.

should read bins=np.arange(nside2npix(2 ** 8) + 1)

constant objects and types may sometimes have a dtype attribute, but no shape.

Replace Plotter(...) by self.class(...) or similar

fname not referenced

Needs to update the key function

The option is parsed but never processed.

It seems dsplotter.corner ignores the labels.

Legends in matplotlib do not include images in general. This should not be an issue because images should always be associated with colorbars.

But there are cases like overplotting multiple densities, one would like to have a "mini colorbar" and a name as a legend.

Would be nice to have a map method to a field of a data container such as in pandas.

df.map('department', get_dep_idnum, ...)

Similar to the norm parser in the plotting functions, it would be nice to have an agg one:
e.g., plot('ra', 'dec', agg='mean(ag)')

such that this finds the function from datashader and does the job.
It should also still allow Agg objects here.

When string lengths are fixed and not identical this breaks

correction

            elif (self.dtypes[name].char == 'S') & (h5obj.dtype.char == 'S'):                                                                                                                                                                 
                dt_size = max(4, max(self.dtypes[name].itemsize, h5obj.dtype.itemsize))                                                                                                                                                       
                self.dtypes[name] = np.dtype('S' + str(dt_size))