MS SQL Server returns two rows for a unique key

A customer has a problem with inserting a certain type of requirements,

and this is because this customer has two keys, TEST-1000 and ??TEST-1000 (? representing an unknown character, possibly an emoji), and MS SQL Server returns both when running the query "KEY" IN ('TEST-1000').

How we’ve resolved it

Upon installing RY 3.4.20, the administrator will be notified if they go to the Requirement Yogi administration that something is not right, and will be encouraged to go to the Requirement Yogi administration in Confluence → Upgrade Tasks → V70 → Details.

Technically

• When installing RY 3.4.20, an upgrade task runs and diagnoses colliding keys (according to the customer’s DBMS).

• If one is found, then a message is displayed in the administration, telling the administrator to go to the Upgrade Tasks → V70 → Details,

• On that screen, the administrator must check the keys which need to be merged, and click “Fix”.

• It is not possible to ignore this required action. So why don’t we automatically fix the keys? Because it merges data, therefore it must be approved by the customer, in case the various keys were actually due to differences. In this situation, the customer must fix their database so that key = ? returns only 1 key at a time.

• Fix in bulk: The “Fix” button also works for an entire page of 100. If this is not enough, then please copy-as-curl (right-click → Inspect → Find the query → copy as curl → paste it in the command line) and change the ‘limit’ parameter.

This kind of issue should not occur during normal operations, since the software checks for the presence of the key before inserting it into the database.