Last modified: 2011-03-13 18:05:30 UTC
CU results, particularly for ranges or IPs with regular editors, can be very long. Right now it's not very flexible and can crash browsers when there is a large result. The results, both for "get edits" and "get users" should be paged like history or contributions pages, i.e. "View (previous 500) (next 500) (20 | 50 | 100 | 250 | 500)"
This could perhaps be done by storing a key with the user id, checked ip/name, and check type for a few hours. The key's existence would allow paging results without adding another log entry. I already have an index enhancement change held up due to the schema change, so I might as well see if I can fit this in.
Could this potentially also incorporate the ability to go back to CU results without actually re-running the check, but instead just accessing a cached copy of the results accessible only to the CU who ran the check? (though, actually allowing any CU to see it would be nice, you can just say "look at these results" instead of copy-paste and emailing or what have you.
This is too inefficient, since it would keep retriggering the same large filesort per page view.