JobQueue observer

[adam-fowler]

Add array of observers to queue. When events like a job is added, finished, failed occur these observers are informed. This allows us to separate out non-core job handling code into individual observers. eg

• publishing events to service eg mqtt, rabbitmq, AWS SQS
• schedule new jobs based on jobs finishing

If we define an observer protocol there are two ways we can define this

1. A protocol that includes methods for every event
2. A protocol that has one function onEvent which is provided with a event type (with additional metadata)

2 has the advantages of being more flexible than 1 but it might have performance issues as every observer would have to get the event type before working out what to do with it.

@Joannis @thoven87

[thoven87]

Add array of observers to queue. When events like a job is added, finished, failed occur these observers are informed. This allows us to separate out non-core job handling code into individual observers. eg

• publishing events to service eg mqtt, rabbitmq, AWS SQS
• schedule new jobs based on jobs finishing

If we define an observer protocol there are two ways we can define this

1. A protocol that includes methods for every event
2. A protocol that has one function onEvent which is provided with a event type (with additional metadata)

2 has the advantages of being more flexible than 1 but it might have performance issues as every observer would have to get the event type before working out what to do with it.

@Joannis @thoven87

I agree with defining a protocol which includes a method for every event is more performant. What's the actual cost of the send options? Since every job would already have all possible observers by default?