Celestial frame hardcoded to 2 axes?

[mdboom]

I don't know if this is a bug or not, but it's come up in the work to serialize frames to/from ASDF.

If I assign a reference_frame on a CelestialFrame to a coordinate that has inherently 3 axes, it doesn't complain, but then len(axes_names) != naxes.

In [17]: import gwcs

In [18]: x = gwcs.CelestialFrame(reference_frame=coordinates.Galactic())

In [19]: x.axes_names
Out[19]: [u'lon', u'lat', u'distance']

In [20]: x.naxes
Out[20]: 2

[nden]

@eteq I am using frame.representation_component_names to get the default names of the frame axes but this includes also distance. What is the correct way to get the default axes names in a frame?

Also gwcs.CelestialFrame.naxes is hardcoded as 2. This isn't always correct (e.g. Galactocentric ). Is there a way to get the number of axes in an instance of one of the builtin_frames?

[eteq]

@nden - if what you want is the names specifically for the default representation (as opposed to the current one for a particular object, which can change at runtime), then this is probably the way to go:

>>> g = coordinates.Galactic()
>>> g.default_representation
astropy.coordinates.representation.SphericalRepresentation
>>> g.representation_info[g.default_representation]
{u'names': (u'l', u'b', u'distance'),
 u'units': (Unit("deg"), Unit("deg"), None)}
>>> g.representation_info[g.default_representation]['names']
(u'l', u'b', u'distance')

Internally, that's basically what representation_component_names does, but it gives you the names for the current representation, which is not necessarily always the default. (But maybe that's actually what you want?)

@nden @mdboom - As for CelestialFrame.naxes, I would think it should be hardcoded to 3. All coordinate frame objects by design are 3-dimensional: they use the representation heirarchy, and currently those are all 3D. UnitSphericalRepresentation is a bit subtle in that it seems to be 2D, but in fact it's interpreted as points on the unit sphere. So you can do this:

>>> s = UnitSphericalRepresentation(10*u.deg, 42*u.deg)
>>> s.to_cartesian()
<CartesianRepresentation (x, y, z) [dimensionless]
    (0.73185479, 0.12904574, 0.66913061)>

And you can see that it's actually 3D.