Ward Fisher

I'll take a look on our end; the rest of my day is spent in meetings, but I'll see what I can figure out.

From a conversation this morning I think we are leaning towards `UNDEFINED` :). Thanks, getting to the PR's as quick as I can :).

This can currently be worked around by passing `--disable-filter-testing` at configure time or `-DENABLE_FILTER_TESTING=OFF` when using cmake.

Note to self: failing at `libsrc4/nc4hdf.c:628`. Need to sort out the context so I can determine if the check that's failing can be modified.

> Update: Not sure this statement is accurate but leaving for future reference for now. The issue appears to be related to an error code thrown by `H5DWrite` at `libsrc4/nc4hdf.c:816`.

Re: the call to `h5Sget_simple_extent_dims` at `libsrc4/nc4hdf.c:613`, ensure what we expect the size of `fdims[1]` to be. Shouldn't it be 4?

Ok, the issue appears to be when writing the file via `ncgen`, the dimension data is being written to `_nc4_non_coord_aName` instead of to `aName`, but `aName` is being queried by...

Using `ncgen` to generate the binary file (as outlined above) from the CDL and then using `h5dump` to inspect the generated file shows results in the observation from 2 days...

note to self: ncdump is working properly; the call to `h5Sget_simple_extent_dims()` (in the 900s) invoked by ncdump returns the values expected, unlike the call referenced above, at 613. See `gdb_ncdump.txt`...

From a conversation with Russ Rew, we are working with **Auxiliary Coordinate Variables**. Here is some information that can be used for determining what _should_ happen: - [Definition of a...