ObsCore 1.1: Erratum 1

Author: DM WG

Date last changed: 2018-11-18

Date accepted:

Rationale

This erratum corrects a few issues noted in the ObsCore-1.1 document . Since most are small typo level items, it we include multiple items in one erratum.

1. Table 6 lists invalid UCD strings for certain elements (invalid with respect to the UCD1+ standard) in obstap columns definition.
2. Implementation feedback from Marco Molinary, see http://mail.ivoa.net/pipermail/dm/2017-August/005604.html

Erratum Content

Invalid UCDs in Table 6

page 56, Table 6: TAP_SCHEMA.columns

• Invalid UCDs
  • obs_publisher_did - meta.ref.uri;meta.curation
    • The problem being that 'meta.curation' is a primary word (P) being used as secondary. 'meta.ref.uri' is indifferent (Q)
  • publisher_id - meta.ref.uri;meta.curation
    • same as obs_publisher_did
  • o_stat_error - stat.error;phot.flux
    • the issue here being that the addition of 'phot.flux' is too specific for the usage.
• Correct UCDs
  • obs_publisher_did - meta.ref.ivoid
  • • 'meta.curation' cannot be preserved because it is a primary word (P) being used as secondary. 'meta.ref.ivoid' is indifferent (Q)
    • ivoid already refers to a curation context
  • publisher_id - meta.ref.ivoid
    • Same as above
      
  • o_stat_error - stat.error

pol_states

• Incorrect text
  • This field is listed as mandatory in §3.2 (Table 1, page 21) but then, Appendix B page 42, in Table 5 the MANdatory column says NO. After that, Table 6 on page 57 lists pol_states again among the mandatory fields. This looks like simply a typo.
• correction
  • Correct setting in Appendix B: Table 5 on page 42, to show pol_states as MAN=YES

t_refpos

• issue
  • This field is listed in Table 5 (Appendix B) page 41 as an optional one, but has no other entry in the specification, e.g. it has no entry in Table 7 (Appendix C.2) so that no Utype or UCD is defined for it. This one looks like a simple forgetfulness.
• answer
  • as we are considering Time series to be handled by ObsCore for discovery, this reference for the time system has to be re worked. Feedback from the dm list is welcome for this
• correction
  • It seems this is a placeholder for future expansion of ObsCore, it seems appropriate to remove this item from the Table 5, and restore it when the full specification can be provided in an update to the REC.
  • t_refpos entry has been removed from table5

units for strings

• issue
  • Table 5 (pages 40-43) reports units for the various fields. However it defines string-type fields to be unitless except for s_region (no value is reported) and proposal_id (which is set as unit=string).
• correction
  • correct both proposal_id s_region entries in Table 5 to reflect unit=unitless

missing facility_name

• issue
  • facility_name missing in table 5
• correction
  • added as mandatory field

Inconsistency in 'principal' settings

• issue
  • The following fields, all declared as optional (with a MAN = NO) in the TABLE 5, have ‘principal’ set to 1 in TABLE 6. It should be 0 for optional fields.
    • dataproduct_subtype
    • o_calib_status
    • obs_creator_name
    • obs_release_date
    • obs_title
    • s_pixel_scale
    • target_class
    • obs_creation_date
    • publisher_id
    • s_ucd
    • s_unit
    • s_resolution_max
    • s_resolution_min
    • em_ucd
    • em_res_power_max
    • em_res_power_min
    • em_resolution
    • o_unit
• correction
  • The meaning of the 'principal' column is explained in C2 page 55. Its is not equivalent to mandatory status of the field. All mandatory fields are principal and some optional fields can be principal either (table7).
  • The definition of the 'principal' column has been updated in P55.

Impact Assessment

Implementations which follow the ObsCore specification for these UDS are compliant with the model spec, but produce a validation error by taplint which validates the UCD value against the UCD specification.

Notes

1. I do not see 'meta.ref.ivoaid' in the UCD list..
2. The proposed changes are in the errata proposed document attached here http://wiki.ivoa.net/internal/IVOA/ObsCore-1_1-Erratum-1/REC-ObsCore-v1.1-20180222_errata.pdf
3. In the above, it is not clear that all errata have been incorporated, needs review by WG.