Behavior of "None" changed in Jira 8.5 and MultiLevel Cascading Select field

[Sean Herring [Kaiser Permanente]]

I recently upgraded to Jira 8.5 from 7.13. 

We use SourceSense's Multilevel Cascading Select field so users can route issues to specific teams using a MLCS field called Routing.

Routing field has 3 levels. For simplicity, I'll call them Region, Department and Team.

We also have the following JQL so users can see if any issues have been routed to their department, but the team hasn't been identified:

Routing in (Region CF ID, Department CF ID, None)

The expected behavior of this would be the query will return all issues where Region and Department have been populated but the Team field hasn't. This is how it worked prior to Jira 8.5.

After the upgrade, this query returns all issues where the Region and Department match, regardless of what the Team option says. This defeats the purpose of this filter. 

I've tried using "Empty" and "Null" in place of "None". I've also tried using various combintion of wildcards and !. Nothing worked. 

Our temporary workaround is to go in and add "Routing not in (Region CF ID, Department CF ID, Team 1 CF ID) Routing not in (Region CF ID, Department CF ID, Team 2 CF ID)..." However, I have a lot of teams and this will require a considerable amount of maintenance. 

Also, I've both emailed and filed a ticket with SourceSense, but they have yet to respond and they don't have a great reputation for responsiveness to begin with.

I've also tried various combinations of Indexing language, stemming and so forth. 

Here are my questions. 

1. Is anyone aware of a change of the behavior of the "none" function in JQL in the upgrade from 7.x to 8.x? 

2. I have Scriptrunner available and am pretty handy with Groovy. Can anyone see (or has seen) a way to reduce the results returned using a Behaviour?

3. Any other ideas/suggestions?