change AstropyWCS2 to correctly expose coordsys as a string - #254

@eteq , my test case was just a header without any WCS info.

I am getting this now for 'astropy2'. With 'astropy', I do not get any traceback. Both versions give me "no WCS" on the display, which is expected.

 File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/misc/plugins/Pan.py", line 249, in set_image
    image.add_offset_xy(x, y, 1.0, 1.0)
  File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/AstroImage.py", line 362, in add_offset_xy
    ra_deg, dec_deg = self.pixtoradec(x, y)
  File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/AstroImage.py", line 310, in pixtoradec
    ra_deg, dec_deg = self.wcs.pixtoradec(args, coords=coords)
  File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/util/wcsmod.py", line 326, in pixtoradec
    return self._frametofloats(self.pixtonative(idxs, coords=coords))
  File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/util/wcsmod.py", line 348, in pixtonative
    self.realize_frame_inplace(sky)
  File ".../ginga-2.5.20151215192336-py2.7.egg/ginga/util/wcsmod.py", line 286, in realize_frame_inplace
    if (issubclass(self.coordframe.representation,

Let me put that file in Box, so you can go grab it and see for yourself.

@eteq, it would be great to have a unit test for this. I've just checked in a wcs unit test file that I've had sitting around for a while. That would be a great place to add a test. You might need to rebase the PR because it has some small changes to wcsmod. The test file is in ginga.tests called test_wcs.py

aha, I think I see the problem, and can fix it in a proximate sense, but before I do, just so I don't induce more problems... @ejeschke, can you clarify for me what the "outside" interface is to the WCS? Is http://ginga.readthedocs.org/en/latest/manual/internals.html#i-want-to-use-my-own-world-coordinate-system still accurate, or is there more since that was made? (coordsys isn't on there, for example.)

And what is the difference between pixtoradec and pixtosystem?

@eteq, I've updated the documentation in the base class for WCS wrappers. Let me know if you need more information.

@ejeschke , can you confirm if updating the documentation triggers automatic RTD build? (for #219)

@pllim, doc builds are now triggered correctly.

Yay! Thanks for checking, @ejeschke ! 👍

I see that there has been some work to optimize AstropyWCS in the Astropy repo. Is AstropyWCS2 still necessary in Ginga?

Yes, and it infact needs a re-write to make use of the new optimisations in 1.3. It's on my todo list, but can't say it's very near the top!

Is the correct WCS handling for your solar frames merged into astropy.wcs, @Cadair ? I think that was another one of the original reasons for this version.

It would be useful to have this version fixed up if it can make use of the new optimizations. The problem caused by the speed of coordinate conversion was fixed with a workaround in ginga, but making use of the optimization would speed up rendering of graphics that are plotted in WCS coords.

@ejeschke astropy.wcs does not do a good job with solar stuff, it get's the base maths right but it doesn't follow convention etc. astropy.coordinates in combination with sunpy.coordinates does all that on top of astropy.wcs though.

I will try and bump this up my todo list.

@Cadaur - do you want to do this in a separate PR (i.e., I should close this), or as an add-on to this or something? At this point it's probably stale enough that maybe starting afresh is going to be needed anyway?

@eteq I will take this and start again probably.