sqlite::savepoint

Creates a named SAVEPOINT on a connection and releases or rolls it back automatically. It also integrates with sqlite::snapshot so callers can capture and re-open snapshots inside nested transactions.

Use cases

• Acts like a “mini transaction” that can be nested under a parent sqlite::transaction.
• Great for retrying a chunk of work inside long-running migrations—rollback on the savepoint without aborting the entire outer transaction.
• Expose the current name via getName() so you can correlate logs with the SQL emitted inside the guarded scope.

Example usage

sqlite::transaction outer{db};
sqlite::savepoint block{db, "apply_patch"};

try {
    run_patch(block.getName());
    block.release(); // optional; destructor releases automatically
    outer.commit();
} catch (...) {
    block.rollback(); // rewinds partial changes
    throw;
}

Pair a savepoint with a snapshot when you need a consistent view before touching the data; it keeps long migrations from surprising the writers.

Public API