JIRA has had, for a long time, a filter to translate certain characters from the CP1252 codepage. These caused issues with MS Word data in JIRA issues displaying incorrectly on non-Windows systems.

There is a bug in the method - StringUtils.escapeCP1252() - that performs this escaping. This bug causes these special characters to be escaped in UTF-8 as well as CP1252. This is documented in CORE-100.

The customer impact of this is that customers cannot use JIRA with Crowd and Active Directory if they have group names that contain em-dash or en-dash. These look like hyphens, but aren't.

Confluence and Crowd no longer use the WordCurlyQuotesFilter that causes this problem. I'd be very happy if JIRA could either a) also abandon it, or b) fix CORE-100.