On Wed, Jul 23, 2014 at 6:18 PM, Philip Bennefall <[email protected]> wrote:
> I have to amend my last message. The timings I just gave was for looking > up that word 10 times, not 1. So the longest time I've seen would be about > 150 ms. However, if you have a document with a few thousand words we would > still be looking at a significant total searching time. Is this to be > expected? > There is no expectation. Spellfix is an experiment in doing fuzzy matching. It was designed for a specific customer who is doing spell-checking in real-time, as the text is being entered. Spellfix works way faster than the end user can enter text, so performance is not an issue in its original purpose. Perhaps you are using spellfix in a different way? You are welcomed to do so. If you want to contribute ideas on how to improve spellfix for use in different scenarios, we will welcome your input. There are comments in the code explaining how spellfix works. Please review the principles of operation and then perhaps run a performance analysis using gprof or cachegrind. Then describe exactly what you are doing and why it isn't working out for you and perhaps we can help. -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
