Point in Coord

[CresitelloDittmar, Mark]

Laurent,

The hierarchy is: Point references a SpaceSys which contains the SpaceFrame
and the CoordSpace (both in composition).
Point
   |-> SpaceSys
         o-> SpaceFrame
         o-> PhysicalCoordSpace{Cartesian/SphericalCoordSpace}

But that does not negate the question.
  "However, taking into consideration that spatial coordinates is the most
used thing in Astronomy, I'm wondering if it wouldn't be better to have one
specific class for Cartesian points (refering to CartesianCoordSpace) and
another for Spherical points (refering to SphericalCoordSpace)."

  * Earlier drafts of Coords (2018) had Frame-centric coords with standard
spaces (CartesianCoord, LongLatCoord)..
  * Due to feedback on this representation, they migrated in 2019 to
specialized singular coordinates (X,Y,Z,Long,Lat,R, etc) which referred to
axes of standard spaces, and were used in frame-centric Measures.  Which is
what went to the RFC phase.
  * There, the frame-centric and space-centric Measures were generally
disliked (GalacticPosition, CartesianPosition)
  * The RFC actions called for replacing the specialized singular
coordinates with a single Point Coordinate, and removing the specialized
Measures, retaining only the single Position type containing a Point.  The
consequence of users having to interrogate the Position to determine the
details of frame/space was considered acceptable.

Obviously there is a sweet spot there somewhere, but I doubt we can settle
into it until we have more implementation experience with it.  Adding a
CartesianPoint and SphericalPoint which constrains the space is a simple
update which can be done at any time.

Mark





On Tue, Apr 7, 2020 at 5:42 AM Laurent MICHEL <
laurent.michel at astro.unistra.fr> wrote:

> Dear DM
>
> I'm exercising with Coords with spatial coordinates.
>
> My understanding is as follow:
> ==============================
> Spatial coordinates are represented by Point instances that, skipping
> the details, refers to a SpaceSys that refer to a SpaceFrame that refer
> to a PhysicalCoordSpace that is either a SphericalCoordSpace or a
> CartesianCoordSpace.
>
> Point
>    |-> SpaceSys
>          |-> SpaceFrame
>               |-> PhysicalCoordSpace{Cartesian/SphericalCoordSpace}
>
> So a client that gets a Point instance will have to step down this
> cascade and to check the class of the associated PhysicalCoordSpace
> instance before to know whether this point is  Cartesian or Spherical.
>
> Question:
> ========
> This is consistent but not very practical. I understand that this model
> provides components for host models that will be designed in a way to
> avoid clients to do such inferences. This could be the case with the
> upgrade of Meas.
> However, taking into consideration that spatial coordinates is the most
> used thing in Astronomy, I'm wondering if it wouldn't be better to have
> one specific class for Cartesian points (refering to
> CartesianCoordSpace) and another for Spherical points (refering to
> SphericalCoordSpace).
>
> As a side effect this would allow to have one specific spherical
> CoordSpace for the celestial sphere (lat ,long, R=1).
>
> Laurent
> --
> ---- Laurent MICHEL              Tel  (33 0) 3 68 85 24 37
>       Observatoire de Strasbourg  Fax  (33 0) 3 68 85 24 32
>       11 Rue de l'Universite      Mail laurent.michel at astro.unistra.fr
>       67000 Strasbourg (France)   Web  http://astro.u-strasbg.fr/~michel
> ---
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20200408/c64e79b4/attachment.html>