It appears that we match word prefixes only (which is arguably a good choice): "ath" does not find "LibreOffice Math", but "math" does. However there are some cases where U8's implementation is annoying.
So Unity7 is returning the more useful result first, but it's still not perfect. I would have said the perfect answer would be to only return "Simple Scan" for "si". If we did that then the alphabetical split would not matter (only one result returned).
It appears that we match word prefixes only (which is arguably a good choice): "ath" does not find "LibreOffice Math", but "math" does. However there are some cases where U8's implementation is annoying.
Example: Search "si"
Unity7 returns: "Simple Scan", "Universal Access", "Disk Usage Analyzer", "Passwords and Keys".
Unity8 returns: "P: Password...", "S: Simple Scan", "U: Universal..."
So Unity7 is returning the more useful result first, but it's still not perfect. I would have said the perfect answer would be to only return "Simple Scan" for "si". If we did that then the alphabetical split would not matter (only one result returned).