Fix: Util function to ensure list of dicts all have same keys for hek

[nabobalis]

Fixes #3238 via a large HACK.

Not really sure this is a sane way to do it however.

[pep8speaks]

Hello @nabobalis! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file sunpy/net/tests/test_hek.py:

Line 172:101: E501 line too long (134 > 100 characters)

Comment last updated at 2019-06-25 19:13:16 UTC

[Cadair]

So, let me check I understand this, some records that are coming back from the HEK have different (in table parlance) columns than others? So you are making sure all the rows contain the superset of all columns?

[nabobalis]

Yes. if the column doesn't exist, we add it and set the value to None.

[Cadair]

I think a different approach would be to use the columns= kwarg to Table, and just pass it a list of the superset of all columns. You could construct the set of all columns by doing set operations on the keys.

i.e.:

>>> d1 = {1:2, 3:4}; d2 = {3:4, 5:6}                                                                                                                                                                                                               
>>> s = set()                                                                                                                                                                                                                                      
>>> s.union(*map(lambda x: x.keys(), [d1, d2]))                                                                                                                                                                                                    
{1, 3, 5}

I can not promise this will work though, but it would mean not having to actually change the list of dicts.

[nabobalis]

K

[Cadair]

Also the constants thing needs a changelog entry of its own.

[nabobalis]

Also the constants thing needs a changelog entry of its own.

Done.

[nabobalis]

Trying to fix that astropy dev failure.

[nabobalis]

What happens if you set the units and constants versions at the start of conftest.py

[nabobalis]

Database failures are back on Linux 3.7. Which is why it seems both the offline and astropy dev build failed.

[nabobalis]

So I changed them to macos.