Why do topics need to be created?

[bdamm]

The protocol requires that publishers and subscribers create topics before publishing or subscribing to topics. There is an option for a single-shot topic creation with publish, but no option for subscribing to a topic that hasn't been created yet.

The result is that if you have a very large tree of potential but not realized topics, subscribers need to actually create a topic before subscribing, regardless of whether the topic will actually be used or not. This converts potential topics into actual topics, which is a major resource burden in the case where there is a large number of potential topics but a small number of actual topics within that potential topic tree.

[jaimejim]

Hi @bdamm , thank you for the comment (and sorry for the very belated response).

The draft has undergone many changes since Janruary, please have a look at the latest version https://core-wg.github.io/coap-pubsub/ps-rev/draft-ietf-core-coap-pubsub.html . In particular the topic lifecycle section: https://core-wg.github.io/coap-pubsub/ps-rev/draft-ietf-core-coap-pubsub.html#name-topic-lifecycle

The draft now splits the part of topic configuration and actual pub/sub data interactions. A topic is not fully created until someone posts in it, thus subscribers can only find actual working topics.

However in any case, for subscribers to subscribe the publisher should have created the topic first. Therefore, on the topic of single-shot subscription to a non-existing topic, the result would be the same as a GET to a CoAP resource that does not exist; 4.04 Not Found.

A broker manager could change this behaviour and we could envision a mechanism by which the CoAP server allows for observations on resources that not-yet exist but I would guess that this is something to be checked with CoRE.

Let core (core@ietf.org) know if you are still interested in this topic please.