Bugfix for empty scheduling block list

[bmorris3]

Hoping this is the last fix blocking the update for astropy 1.3 compatibility, which affects #296, #292, #282.

[bmorris3]

Hey @eteq – I have a question for you about the implementation of sites within EarthLocation in astropy. It looks like when setup.py test is run on astroplan without --remote-data, astropy is installed without remote data, and therefore the only site in the registry is Greenwich. Previously, astroplan had monkey-patched APO, Subaru, and maybe other sites into the Observer.at_site method so that we could still run some doctests even without --remote-data.

It looks like that's now out of date, and this issue is the last blocker that we need to fix in order to release a much needed bugfix update for astroplan.

Do you have thoughts on how best to approach this problem? Would you accept a pull request to sites.json that includes the site information for the sites that astroplan tests on, or should we re-work those astroplan tests away from real observatories (like this one that uses Subaru, for example)? Or we could re-write the monkeypatch for astroplan (this time onto EarthLocation) just for those sites that we use in the astroplan tests.

Perhaps there's already a patch being done here, and the patch is failing for reasons I don't understand (failure error here)?

cc @bsipocz – we're so close to being ready for that bugfix release.

[bsipocz]

@bmorris3 - this seems to be very similar to the issue we have currently in astropy, e.g. https://travis-ci.org/astropy/astropy/jobs/221883232#L2337

I suggest to use a local monkeypatch rather than to add a few random obs to astropy's default. The latter will raise questions about the selection of obs getting into that file while others are left out and we've already been down that road before.