DC2C migration - discrepant filter results using "text~'two words'"

[Trudy Claspill]

Hello folks,

I have executed a DC to Cloud migration for Jira Software DC v9.10.1 using JCMA 1.9.17.

During the validation immediately after the migration, before users have accessed the system and started changing data, I have found that filters that used a criteria like

text~'two words'

...are returning different results sets in the source Jira DC instance vs. the destination Jira Cloud instance. In most cases the Jira Cloud results show fewer issues.

I have reviewed the documentation for using "text~" in Jira DC.
https://confluence.atlassian.com/jirasoftwareserver0910/search-syntax-for-text-fields-1282248659.html
That documentation says specifically when the multiple words are contained in quotes for searching the text master field, then the results will return issues that have both words in no particular order in the support text type fields.

And I have reviewed the documentation for using "text~" in Jira Cloud.
https://support.atlassian.com/jira-software-cloud/docs/search-for-issues-using-the-text-field/
I didn't see where it said anything about how multiple words are handled in the text search.

In Jira Cloud I have looked up the issues returned by the Jira DC filter. They exist and they still meet the criteria of the filter based on the assumption that he results will return issues that have both words in no particular order in the support text type fields.

I've looked for information related to this on the internet, in the community, and in Atlassian's public backlog, but haven't found anything. I've already opened a support ticket but thought I would also post a question here.

I'll respond with what I learn from Atlassian support.

[Walter Buggenhout]

Hi @Trudy Claspill,

Maybe this Community article on searching for an exact phrase may be relevant. It is not an exact match with the behaviour you describe, but I can imagine the changes they made impact your scenario as well. 

[Trudy Claspill]

Thanks @Walter Buggenhout 

The option to do an exact phrase match was available in Jira DC in addition to the non-exact phrase match that we use. We want the non-exact phrase match, where the search results return issues that have all of the words specified but in any order.

With additional experimentation I found that if I call out the individual text fields explicitly rather than use text, then I can get the same results in Jira Cloud as in Jira DC. 

Example:

summary~'two words' or description~'two words' or comment~'two words'

But that is not a sustainable solution - keeping track of all new qualifying custom fields and adding them to the filters explicitly.

Text

This is a master-field that allows you to search all text fields for issues, such as:

• Summary

• Description

• Environment

• Comments

• custom fields that use the "free text searcher"; this includes custom fields of the following built-in custom field types:

  • Free text field (unlimited text)

  • Text field (< 255 characters)

  • Read-only text field