Optimistic Concurrency

Two users load the same row, both edit it, both save. With no protection the second save silently overwrites the first. Optimistic concurrency control (OCC) catches that.

How it works

Give the table an integer version column. Pika scopes every update to the version it read:

UPDATE posts SET ..., version = 2 WHERE id = ? AND version = 1

If someone else changed the row first, its version is already 2, so WHERE version = 1 matches nothing and the update touches 0 rows. Pika sees that and update() returns false. (Through saveOrThrow(), a failed save throws IllegalStateException.)

Turn it on

Add a field named version. That alone enables OCC for the class.

public class Article extends PikaBean {
    long id;
    String title;
    Integer version;   // the presence of this field turns on OCC
}

Pika sets version to 1 on insert and bumps it on every update. Handle a conflict by checking the return value:

Article a = Article.find().byId(1L);
a.setTitle("edited");
if (!a.update()) {
    // someone changed this row first: reload and retry, or tell the user
}

Customize the column

For a different column name everywhere:

new PikaORM("jdbc:sqlite:app.db")
    .withDefaultVersionColumnName(clazz -> "opt_lock")
    .makeDefaultORM();

Or per class, in the mapping:

@Override
public FieldMapping mapField(Field field) {
    if (field.getName().equals("optLock")) {
        return map(field)
                .toColumn("opt_lock")
                .asVersionColumn()
                .withVersionIncrementer(v -> v == null ? 1 : ((Integer) v) + 1);
    }
    return defaultMapping(field);
}

Turn it off

If you have a version field that means something else (an API version, say), opt out so Pika leaves it alone:

new PikaORM("jdbc:sqlite:app.db")
    .withNoDefaultVersionColumn()
    .makeDefaultORM();