Release notes for Soda Cloud Reporting API
Review release notes for the Soda Cloud Reporting API.
Last updated
Was this helpful?
Review release notes for the Soda Cloud Reporting API.
Last updated
Was this helpful?
Was this helpful?
09 November 2023
Bug fix following the removal of creator_
fields causing SQL query errors for check_results
and dataset
endpoints.
08 November 2023
Fields related to creator information (with creator_
prefix) are now deprecated from the /checks
, /check_results
, datasets
endpoints. Those fields were legacy from soda-sql. With soda-sql data removed from our system the concept of creator
is also removed from the reporting API. Use the owner_
fields instead.
07 November 2023
The Soda Cloud Reporting API v0 is now deprecated. Please use Reporting API v1.
09 June 2023
Released as generally available.
Read more in the Reporting API Changelog and v1 migration guide.
15 December 2022
Released in Preview state, version 1 offers improvements and additions to Soda Cloud’s Reporting API endpoints.
Better performance as the API retrieves all of its data from pre-processed and aggregated data sources.
“Tests” have been renamed “checks”.
Most endpoints now use pagination.
A few endpoints have been deprecated, a new one, Incidents, has been added, and some have changed for the better.
Read more in the Reporting API Changelog and v1 migration guide.
28 July 2022
The datasets
endpoint will not return the is_deleted
on any of the dataset records as this field has been deprecated by the backend and was never a good indicator of dataset deletion due to hard-deletes.
21 April 2022
The datasets
endpoint now provides creator_info
and owner_info
along with number_of_rows
.
19 April 2022
The reporting API now accepts Soda Cloud’s API keys via headers API_KEY_ID
and API_KEY_SECRET
to allow users with SSO to use the API.
01 April 2022
Users must now query the tests_and_results
endpoint with a single dataset_id
. Documentation reflects this to users and the API throws the errors.
Users must now query the dataset_health
endpoint with a single dataset_id
. Documentation reflect this to users the API throws the errors.
Endpoint usage is now tracked in segment. Every time you send a query to any of the end points, the Reporting API pushes an event that contains the following information:
user_id
organization_id
endpoint_name
environment