Currently, if the namespace list used to sort files result in many files in the same sort bucket, those files are displayed in seemingly random order.
Example: if I have my sort order set as "volume-page" and I have 3 files:
file1
volume:1
file2
volume:1
file3
volume:1
page:3
When they are displayed, file3 is guaranteed to come after file1 and file2, but file1 and file2 are in the same sort bucket and neither one is guaranteed to appear before the other.
I think the sorting should not allow more than one file per bucket, and should fall back to using the file hash as the sort key of last resort should the namespaces be insufficient to sort to one file per bucket.
This will mean that even if files are poorly tagged, their display will still be consistent, which will make it easier to find files by memory.