You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
/CHANGES/CHANGE/WHERE/SECTION and /CHANGES/CHANGE/WHERE/LABEL can refer to non-existent sections or labels.
/CHANGES/CHANGE/WHERE/SECTION and /CHANGES/CHANGE/WHERE/LABEL can refer to existing sections and labels, but can be incorrect.
Some elements on other forms are unlabeled, so /CHANGES/CHANGE/WHERE can't deterministically identify the element to update.
The type of /CHANGES/CHANGE/NEW_VALUE can be inconsistent with the type of the element (e.g. TEXT for a CPV code).
- /CHANGES/CHANGE/NEW_VALUE/TEXT can be incomplete (e.g. a description of changes, rather than a new copy of the updated text).
Discussion
The options are:
Where possible, follow the current approach (i.e. overwrite fields). If not possible due to above issues, describe the changes in new fields in the Amendment object.
Always describe the changes in new fields in the Amendment object.
Regarding (1), I'm concerned about an inconsistent approach. In particular, if there are multiple F14 notices for a contracting process, it will be very confusing to a user if the same field is updated sometimes in-place and sometimes in amendments.
As noted by @ColinMaudry, our approach would ideally be informed by actual error rates. For example, if error rates are low, (1) would have better data quality with only few inconsistencies in approach.
Proposal
Until we get data on error rates, we should take the conservative approach (2).
We should also indicate that publishers should describe the use of Amendment objects in their publication policy (and share the link to the template). (This policy should also link to the OCDS for EU profile, and identify the version in use.)
As a separate note, we should indicate that, if an OCDS release exists for the given ocid, that mapping sections I and II on F14 is optional.
The text was updated successfully, but these errors were encountered:
Background
- /CHANGES/CHANGE/NEW_VALUE/TEXT can be incomplete (e.g. a description of changes, rather than a new copy of the updated text).
Discussion
The options are:
Regarding (1), I'm concerned about an inconsistent approach. In particular, if there are multiple F14 notices for a contracting process, it will be very confusing to a user if the same field is updated sometimes in-place and sometimes in amendments.
As noted by @ColinMaudry, our approach would ideally be informed by actual error rates. For example, if error rates are low, (1) would have better data quality with only few inconsistencies in approach.
Proposal
Until we get data on error rates, we should take the conservative approach (2).
We should also indicate that publishers should describe the use of Amendment objects in their publication policy (and share the link to the template). (This policy should also link to the OCDS for EU profile, and identify the version in use.)
As a separate note, we should indicate that, if an OCDS release exists for the given
ocid
, that mapping sections I and II on F14 is optional.The text was updated successfully, but these errors were encountered: