Unexpected result excluding labels

[Tom Claessens]

Situation

I'll soon be migrating one project to another, to clean up the amount of projects we have in our current JIRA setup.

The project everything gets migrated to, already has a few boards, based on a filter that looks like this:

project = ASWR AND "IRD - Dev Team[Select List (multiple choices)]" = Polaris

There are two teams using the same backlog but only see their assigned tickets on their own boards with this filter. The migrated tickets will not have a dedicated team assigned, but can be identified by the labels field.

So my assumption was that adding the additional filter portion below to the already existing filter, would exclude these tickets.

AND labels != "java-service-upgrade"

Expected result

Since the original board filter returned 557 tickets, and there are no tickets yet with a 'java-service-upgrade' label in the project, that the amount of returned tickets remained  the same.

Actual result

The result however was a drop in returned tickets to 335 tickets. Which of course was unexpected, at least for me.

Fix

After tinkering with the setup, I managed to get the desired result

project = ASWR AND "IRD - Dev Team[Select List (multiple choices)]" = Polaris AND (labels != "java-service-upgrade" OR labels is EMPTY) ORDER BY Rank ASC

Question:

It seems that the tickets without a label, also got excluded when I added the not. So I have it working for our purposes, but wonder why tickets with an EMPTY labels field were also excluded.

[Nic Brough -Adaptavist- · Accepted solution]

It's a logically correct, but very counter-intuitive thing for us humans.  We're a bit more clever than computers and make helpful logical deductions automatically which save time, but are not strictly correct.

This is a bit easier to show as an example and some careful phrasing:  Let's say you have three issues, labelled as:

• ABC-1: red, green
• ABC-2: red
• ABC-3: <no labels>

We ask the question "does this issue have a label called 'green'"?:

Human replies:

• ABC-1: Yes, because I've looked at the list and there's a green there
• ABC-2: No, because I've looked at the list and there's no green there
• ABC-3: No, because I've looked at the list and there's no green there

Logical (computer) replies:

• ABC-1: Yes, because I've looked at the list and there's a green there
• ABC-2: No, because I've looked at the list and none of the entries are green
• ABC-3: I do not know because there is no list to look at to see if it has green in it, so I'm not going to report this issue as a match either way.

When you ask the question "is empty", the computer can now answer yes or no because you're deliberately asking for "no list there", which it can answer!

Totally counter-intuitive, but worth bearing in mind when querying list data, until Atlassian make it intuitive (if ever)