Some of our transactions insert many events, and it's inefficient for each query to do two round trips to the server to prepare the statement then execute it. Mysql docs recommend inserting using a single INSERT with multiple VALUES tuples to speed up inserts.
I think we can achieve this by introducing a manyInserter type:
type EventToInsert struct {
ForeignID string
Type reflex.Type
Metadata []byte
}
type manyInserter func(ctx context.Context, tx *sql.Tx, events []EventToInsert) error
and an API to insert many events:
func (t *EventsTable) InsertMany(ctx context.Context, tx *sql.Tx, events []EventToInsert) (NotifyFunc, error) {
// Check noops...
err := t.manyInserter(ctx, tx, events)
if err != nil {
return noopFunc, err
}
return t.notifier.Notify, nil
}
We can provide a default implementation like we do for inserter.
Some of our transactions insert many events, and it's inefficient for each query to do two round trips to the server to prepare the statement then execute it. Mysql docs recommend inserting using a single
INSERTwith multipleVALUEStuples to speed up inserts.I think we can achieve this by introducing a
manyInsertertype:and an API to insert many events:
We can provide a default implementation like we do for
inserter.