So when that happens, everything will be stored in one e-mail per file, and that will solve this problem. In another bug, it's mentioned that Thunderbird will switch to a maildir format, just not as soon as 2.0.

Seems to me that Spotlight is hardwired for paths (1 to 1), and doesn't accommodate databases (1 to many) very well. we can do that easily on our end of course, but I can't tell at first glance how to tell Spotlight that the search term(s) is in messageX, and then how to associate the 'opening' of that virtual reference to the TBird db, *and* have TBird fire up and select said message. What is not obvious to me is how to disassociate the individual message in the flat-file, from the flat-file itself. the GetMetadataForFile('INBOX') (for example) would then feed Spotlight with the new message(s) index data, and a Spotlight query for a term in that message would register a hit with our plugin. if we register the plugin for the 'file type' (make one up if we have to) for the TBird message store files, said plugin will be called each time one of those files (mailbox folders) is updated. Registration of a spotlight meta-data plugin for TBird's flat-file mail message database is fairly straightforward.