Does operator uid() have to be unique across all jobs?

[John Smith]

When setting uid() of an operator does it have to be unique across all jobs
or just unique within a job?

For example can I use env.addSource(myKafkaConsumer).uid("kafka-consumer")
in another job?