I've noticed that in a file search (CTRL+H) in Eclipse, the first time you search against a "Working Set" it takes a considerable amount of time for the search to complete. Subsequent searches on that Working Set run much faster, presumably because there is some sort of index it is searching against rather than the actual files. However, after a (un)certain amount of time that indexing seems to expire and the search routine reverts back to the full file parsing.
What is the lifetime of this cached indexing? Can it be increased or persisted?