Right now if you add characters to a plain text search, the entire database is queried again. It would be much more efficient to only start searching from the places where a substring was already found.
Implementation notes:
To support backspacing efficiently, we should probably keep all the results from the first query and a stack of successive superstrings with indices pointing to the original results.
Adding prefixes should be supportable by starting a search from offset - N where N is the number of prefix bytes.
Right now if you add characters to a plain text search, the entire database is queried again. It would be much more efficient to only start searching from the places where a substring was already found.
Implementation notes:
offset - N
where N is the number of prefix bytes.