> Simon Slavin, Thank you for your suggestion. Our deduper prototoype > uses fuzzy matching methods such as the Levenshtein Distance to > detect duplicates. We have found that these fuzzy matching methods > are best implemented in C++ for processing time requirements. > We would still like to know your experience with SQLite > WAL databases compared to SQlite non-WAL databases. Particularly, we > are in the sqlite read processing in SQLIte WAL databases. Is > possible to SQLiTe WAL databases to have faster read processing than > SQLite non-WAL databases. If so, what method to use to gain the read > improvement? Thank you.
It is possible that you would see the biggest improvement by implementing your matching method in a plain C SQLite extension. Doing would keep moving/converting data back and forth to a minimum as the workload would be made as close to the engine as possible. I mail you a download link to an extension offering a Unicode-aware fuzzy compare function (Levhenshtein-Damerau exactly). Have a look at it, play with it to see how it can fit part of your bill and adapt the code as you whish. Like Simon, I think you should get rid of journaling in your case. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users