It would be nice to have the ability to use a Lucene RAM index which gets constucted from an existing Lucene file based indexed. For example in an JMS setup the master could create a file based Lucene index, share it out to the slaves which in turn use this file based index to populate a RAM index. This would give you the best of two worlds.
Not sure how hard it would be to implement this in an unclustered environment.
[*] read from FS at startup (optionally, should be configurabile)
[*] write back to FS at shutdown (optionally, should be configurabile)
[*] document the behaviour, especially warning about this case: some setups are doing local clustering: 2 instances sharing the directory. This obviously can't work, so they should avoid using this feature of writing back to FS.