Update Stomp Client Recommendation in Docs [SPR-15624]

[spring-projects-issues]

Gunnar Hillert opened SPR-15624 and commented

In the reference guide we are referring multiple time to an outdated and unmaintained version of the Stomp JavaScript client stomp.js:

https://github.com/jmesnil/stomp-websocket

Which library would we recommend going forward?

It looks like there are a few competing forks (no clear winner) around:

https://github.com/JSteunou/webstomp-client
https://github.com/stomp-js/stomp-websocket

There is also an open question on this on Stackoverflow:

https://stackoverflow.com/questions/41119274/stomp-js-is-out-of-date-and-no-longer-maintained-which-stomp-client-to-use-wit

This may also raise a general question on the State of Stomp as a preferred means for doing websocket-based messaging.

Also, how does it relate to our Reactive WebSocket Support in WebFlux?

How does it relate to #19101

---

Affects: 5.0 RC1

Referenced from: commits 8670411, 513461d, 357fb48

Backported to: 4.3.14

[spring-projects-issues]

Rossen Stoyanchev commented

Thanks ghillert and sorry for the slow response.

Indeed the state of things around stomp.js isn't ideal. On the other hand STOMP is quite simple from a client perspective, and not changing that much, so while the forked state is unfortunate, those forks I believe are fairly active, so it shouldn't be a show stopper in practice.

As for STOMP and WebFlux that is still the plan to do along with SockJS just not in 5.0 since it is a major effort and just like with Spring MVC there will be very little or no shared code. We could also revisit our present thinking and experience with STOMP but since SockJS continues to support text messages only, STOMP is still a very strong candidate. I don't think the situation with stomp.js changes that much.

[spring-projects-issues]

Rossen Stoyanchev commented

Resolved via 357fb4.