Before Jaikoz can edit files, they must be loaded into the program?s working memory. Default memory settings will limit how many files can be analyzed at one go. Typically, this might be several thousand files up to maybe ten thousand files (depends on the files?sparsely tagged files don?t require as much space in memory). One option is to work in batches, tagging your collection in chunks. This does make it harder to find duplicates, if dupes don?t load up in the same batch (e.g., some songs sorted to B for ?Beatles?, and others to T for ?The Beatles?). If files are renamed and resorted after tagging (e.g., all songs now consistently grouped under ?The Beatles?) you can identify dupes in a second pass through your collection. This doesn?t resolve the issue for songs present on single-artist releases vs. various artist releases.
A more convenient fix is to increase Jaikoz? memory settings. The online help gives OS-specific instructions in section 3.5. Two forum threads are also helpful:
http://www.jthink.net/jaikozforum/posts/list/776.page
http://www.jthink.net/jaikozforum/posts/list/851.page
In the end, the limit of how many files can be loaded depends on the machine and available RAM. I?ve run Jaikoz with fifty thousand songs loaded. Be patient and save often is all I can say about that. Limits on time and attention may have you going back to working in chunks, albeit larger chunks than with the default memory settings.
One last comment- it is possible to drag and drop from another program to load files into Jaikoz*. So, you can use the sort and filter capabilities of your music player to create a custom load for Jaikoz. For example, I have iTunes create a smart playlist of all tracks that have a blank album field. I then drag that into Jaikoz and see if Jaikoz can find the missing information. By combining several such playlists, you can maintain a Jaikoz to-do list that updates itself.
*- Apparently there is a bug in the Mac version of iTunes that makes dragging to Jaikoz difficult. Paul provided a workaround that is slow, but works.