#36: make configuration value url optional

[turing85]

Resolves #36

[middagj]

Pretty big, will review it later next week

[turing85]

It looks actually bigger than it is. Most changes (about 100 file changes, give or take) are due to additional test modules, but the tests are mostly the same, with only minor changes.

I also took the liberty and did some general cleanup, e.g. replacing field- with constructor-injection where possible, some renamings for clarity, ...

[middagj]

I checked the source code and put a few comments and skimmed the tests. If there is anything of particular interest in the tests, let me know.

@ppalaga Can you check if this would solve your issue?

[turing85]

I checked the source code and put a few comments and skimmed the tests. If there is anything of particular interest in the tests, let me know.

Only important question is: is the amount of tests (and especially the amount of native image builds) okay? Right now, we are building six native images, a build runs for ~30 minutes.

[ppalaga]

@zhfeng could you please check if this solves apache/camel-quarkus#2857 for us?

[middagj]

Only important question is: is the amount of tests (and especially the amount of native image builds) okay? Right now, we are building six native images, a build runs for ~30 minutes.

That is really the downside of native builds. I don't know if there is an option to only do it once discussion is over.

[middagj]

Thanks again! I leave it open to get feedback from Camel people, but looks good to me.

[zhfeng]

Thanks @turing85 and I think it is good to resolve apache/camel-quarkus#2857.
So if there is a empty config, no ConnectionFactory bean will be generated and healthcheck will return 'UP'. Is it right? Aslo I think we need to update https://github.com/quarkiverse/quarkus-artemis/blob/main/docs/modules/ROOT/pages/index.adoc to add a section to explian the multiple named connection factory configuration and some notes.

[turing85]

Thanks @turing85 and I think it is good to resolve apache/camel-quarkus#2857. So if there is a empty config, no ConnectionFactory bean will be generated and healthcheck will return 'UP'. Is it right?

This is partially correct. Basically, if we load the extension without any configuration, no injectable bean will be created, and the health endpoint will return UP, but the corresponding health check will not be loaded (see here). If we still want the health check to be loaded, then this should be an easy fix. But the check will not pick up the beans produced outside the extension. It would be possible to pick them up, however, we would have no chance of giving them meaningful names.

[zhfeng]

OK, I get it. In camel-qurakus-jms, we have a use case to create a custom ConnectionFactory such like

@Named("custom")
ConnectionFatory createCustomConnectionFactory() {
...
}

and to use it in the camel routers. So if we can find a way to enable the health check in this scenario, it would be great. I'm not sure what is about the meaningful names?

[turing85]

The health of those connection factories are represented as key-value pairs in the health check. The key is a display name, the value is it status (see e.g. here). In the example you gave, the name is "custom" (I'd btw recommend to not use @Named, but other annotations like @Identifier or a custom scope). It would be possible to get all instances of the ConnectionFactory-interface, but we are not (easily) able to detect the annotation that qualifies the name.

[zhfeng]

Thanks for explainations. Can we create a follow up issue to track this improvement? and all of others are good to me. Thanks for your contribution again!

[turing85]

@zhfeng I have created #84 to track the follow-up.

On a sidenote: I took another look at the source code of the pooled-jms extension, in particular PooledJmsWrapper, line 44. Am I correct that, for XA-enabled connection factories, connection pooling is not enabled? Scratch that. It was a brain fart.

[zhfeng]

No, I think the pooling is still enabled and just add support for transaction manager integration.

[turing85]

Yeah, I saw the call to pooledJmsRuntimeConfigureConnectionFactory(...) after opening the code in an IDE :) As I said: Brain fart