I just switched to 3.3.10 and I'm playing around with sqlite3_prepare_v2. I have a database with two open connections, and each connection creates its own table. After the second connection creates its table all statements created with sqlite3_prepare_v2 on the first connection fail with SQLITE_SCHEMA. I searched the mailing list and found this earlier discussion about schema changes from different connections: http://www.mail-archive.com/sqlite-users@sqlite.org/msg17823.html, but I'm wondering if sqlite3_prepare_v2 should try to reload the schema before returning the SQLITE_SCHEMA error. The new documentation states:
"If the schema has changed in a way that makes the statement no longer valid, sqlite3_step <http://sqlite.org/capi3ref.html#sqlite3_step>() will still return SQLITE_SCHEMA. But unlike the legacy behavior, SQLITE_SCHEMA is now a fatal error. Calling sqlite3_prepare_v2<http://sqlite.org/capi3ref.html#sqlite3_prepare_v2>() again will not make the error go away." which implies that SQLITE_SCHEMA is only returned when the schema change is fatal, i.e. the table has been removed. If v2 statements are going to continue returning an error for schema changes from another connection, can there at least be separate error codes for fatal vs. non-fatal schema changes?
