There seems to be a disconnect between the queries used to create the counts, and the queries expressed in the URL links.
Here is a rundown of the search URLS and count queries:
Open:
URL: all bugs with unresolved status, omitting dupes
Count: same.
Assigned to me:
URL: bugs assigned to user with unresolved status, omitting dupes
Count: same.
Critical:
URL: bugs with unresolved status, severity=Critical, omitting dupes
Count: bugs with any status, severity=Critical, omitting dupes
Unconfirmed:
URL: bugs with status=Unconfirmed, omitting dupes
Count: same.
Unassigned:
URL: bugs with unresolved status, assignee=NULL, omitting dupes
Count: bugs with any status, assignee=NULL, omitting dupes
All bugs:
URL: bugs with unresolved status (default when none given), omitting dupes
Count: all bugs except dupes.
So the "Critical", "Unassigned" and "All bugs" items can all give misleading counts. In the first two cases, the count should be adjusted. In the last, the URL should be.
It is likely that bug 34224 has the same cause, but I haven't checked the templates to verify.
There seems to be a disconnect between the queries used to create the counts, and the queries expressed in the URL links.
Here is a rundown of the search URLS and count queries:
Open:
URL: all bugs with unresolved status, omitting dupes
Count: same.
Assigned to me:
URL: bugs assigned to user with unresolved status, omitting dupes
Count: same.
Critical:
URL: bugs with unresolved status, severity=Critical, omitting dupes
Count: bugs with any status, severity=Critical, omitting dupes
Unconfirmed:
URL: bugs with status=Unconfirmed, omitting dupes
Count: same.
Unassigned:
URL: bugs with unresolved status, assignee=NULL, omitting dupes
Count: bugs with any status, assignee=NULL, omitting dupes
All bugs:
URL: bugs with unresolved status (default when none given), omitting dupes
Count: all bugs except dupes.
So the "Critical", "Unassigned" and "All bugs" items can all give misleading counts. In the first two cases, the count should be adjusted. In the last, the URL should be.
It is likely that bug 34224 has the same cause, but I haven't checked the templates to verify.