🐛 Bug

Compiling against the C++ API on macOS using GCC-9.3, and cmake seems to use a bad flag:
... -fopenmp -D_GLIBCXX_USE_CXX11_ABI= -std=c++14 ... -- note how it "blanks out" the _GLIBCXX_USE_CXX11_ABI variable. This causes the compiler to fail in the stdlib:

/usr/local/Cellar/gcc@9/9.3.0/include/c++/9.3.0/x86_64-apple-darwin18/bits/c++config.h:273:27: error: #if with no expr

Improve navigation of the I/O how-to by prepending a decision tree/flowchart using an image or graphviz, as suggested by @mattip.

Like xarray, pandas supports attaching arbitrary metadata to DataFrames and persisting it across operations. https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.attrs.html

Dask could pretty easily implement this as well. We'd have a

• _Frame.attrs property. This would likely return self._meta.attrs.
• We'd verify that dd.from_pandas(data) correctly extracts attrs from `da

Reporting a bug

• [x]

Well, Gumbel Distribution is magical. Basically, given a sequence of K logits, i.e., "\log a_1, \log a_2, ..., \log a_K" and K independent gumbel random variables, i.e., "g_1, g_2, ..., g_K". We have

\argmax_i \log a_i + g_i ~ Categorical({a_i / sum(a)})

This gives you a very simple way to sampl

Support Series.median()

What happened:
xr.DataArray([1], coords=[('onecoord', [2])]).sel(onecoord=2).to_dataframe(name='name') raise an exception ValueError: no valid index for a 0-dimensional object

What you expected to happen:

the same behavior as: xr.DataArray([1], coords=[('onecoord', [2])]).to_dataframe(name='name')

Anything else we need to know?:

I see that the array after the select