sometimes i think a more structured cache (not just random global vars) may be a benefit.
I got to think about it when we cache info regardless of access controls, sometimes its different "as seen from a reporter", than it's to core system functionality.
Sometimes we would want to perform operations with contolled/disabled cache (mass treatment of data, one pass)
...etc

I had a quick look, and it appears that user_cache_database_result() is only called after user_search_cache() failed to return anything, which implies that the cache is not set for the record being added.

That being said, I don't see the rationale behind the function's current behavior either.

I stumbled too often over this function and did not understand what it does.

PR https://github.com/mantisbt/mantisbt/pull/866