Error: Is AbstractJmsChannel an abstract class or not?

[dr_pompeii]

Hello

I am doing a deeper analysis for the SI 2.1.1 API and I did realize the follow for the AbstractJmsChannel's API
Where appear

public class AbstractJmsChannel extends AbstractMessageChannel

If it is really an abstract class then should be

public abstract class AbstractJmsChannel extends AbstractMessageChannel

If It is a bug (seems yes) let me know to create a JIRA issue

Kind Regards

[oleg.zhurakousky]

Yeah, this seems like leftover after refactoring. I would not call it a bug, since most likely it was left intentionally since changing class name would constitute a breking change and we try not to introduce breaking changes until the next major release.

[Mark Fisher]

Oleg, do you have the diff handy that shows that refactoring? I don't think we ever made a change where the abstract base class was intended to be used as a concrete channel. In fact, I'm guessing that we never had the abstract keyword on this class, and that as Amol pointed out, it should be there.

[oleg.zhurakousky]

The last change I see is:

commit 55d9366c4b42a2a2f8724f8ed4debc3cb17e0200
Author: Mark Fisher <. . @. .com> 2010-09-23 04:49:39
Committer: Mark Fisher <. . @. .com> 2010-09-23 04:49:39
. . .

And yes you are correct it seems to have always had "Abstract" in its name but was never 'abstract' class. So that's a bug, but I am wondering if we are stuck with it till 3.x since it would be a breaking change.

[Mark Fisher]

It would only be a breaking change if someone is using AbstractJmsChannel directly as a concrete class, and I doubt that's the case. Our two JMS-based channels created by parsers definitely do not.

Therefore, I'd say it's a case where bugginess outweighs any breaking change. I think we should target 2.2. It's basically harmless (just wrong), so it's probably okay to *not* backport this one to 2.1.

Agree?
-Mark

[oleg.zhurakousky]

Yep.

Manuel, you can raise a JIRA

[dr_pompeii]

Hello Oley and Mark

Oleg:

Manuel, you can raise a JIRA

Done: https://jira.springsource.org/browse/INT-2540

Mark:

I think we should target 2.2. It's basically harmless (just wrong), so it's probably okay to *not* backport this one to 2.1.

Is possible know when SI 2.2 would be available?

I did realize SI 2.1 has support about AMQP within (MessageChannel) where I think It was not available in SI 2.0, my point is, I dont want in a good sense spend some time researching with SI 2.1 and later have strong changes about the API for 2.2.

Kind Regards