SQLite Forum

Integer overflow risk in sqlite3_changes and underlying counter?
Looking at [the remedy](https://www.sqlite.org/src/info/48fdec22c966003f), there seem to be some omissions which will continue to produce misleading results on 32-bit architectures when many changes have occurred.

1. In src/func.c, the SQL total_changes() function should be updated to call sqlite3_total_changes64() and return an int64. The SQL changes() function was updated so they are inconsistent as of this check-in.
2. In src/shell.c the ".changes on" display (implemented in the function runOneSqlLine) should be calling sqlite3_changes64() and sqlite3_total_changes64() and printing 64-bit integers.