@PostConstruct deprecated?

[cdprete]

Hello.

I've read on https://www.linkedin.com/posts/gureev-roman_springboot-java-javadev-activity-7345764962612436992-c4bQ?utm_source=share&utm_medium=member_android&rcm=ACoAABZtGW0BD5hF4YhvmOfwFwVPw4kYJnjiMwY that @PostConstruct is now deprecated, but I can't find anything about this in the official documentation.

Is it really so? And, if it's true, can you please reflect it in the documentation?

Since this seems fishy to me, I would like to get your opinion as experts in the matter. 🙂

Kind regards and thanks.

[philwebb]

It's a framework feature, but as far as I know it isn't deprecated. Documentation for the feature still exists. You should be using jakarta.annotation.PostConstruct and not javax.annotation.PostConstruct, perhaps that's caused the confusion.

[cdprete]

Thanks for the reply @philwebb.

I'm well aware that the Jakarta annotations should be - since years - used instead.
To my knowledge, the javax annotations don't even work anymore in Spring Boot 3.4.x.
I claim this because I had to explicitly migrate an internal library to them to make it work again after migrating the a project using it to version 3.4.5.

[cdprete]

Hi @philwebb.

May you share your view about the last point?

To my knowledge of IoC containers, I would be expecting the 2 approaches to be identical.
I fail to imagine a scenario where the @PostConstruct is called before the bean is fully initialized.
Even by throwing in AOP, if applied consistent to both the approaches, it should lead to the same error in both, so I'm seriously failing to understand that point.

Kind regards and thanks.

[bclozel]

Hi @cdprete ,
It is challenging to discuss this matter with an incorrect LinkedIn post and some statements describing an application from your side.

I think you should create a new issue in the Spring Framework project with a minimal sample application that demonstrates the problem: something that works as expected with Spring Boot 3.3.x and breaks with 3.4.x. We can discuss core container behavior from there. Thanks.

[cdprete]

Hello @bclozel.

The point is that I fail to even imagine how things can fail to begin with.
Therefore, it's for me impossible to provide an example and that's why I was asking the opinion of your team regarding such last point around AOP, with the assumption that your team knows the inside-outs of the framework and can, indeed, indicate if such issue described in there happen or not.

[bclozel]

From what I understand, you had a working arrangement previously and you had to change something to make it work with Spring Boot 3.4 (use javax or Jakarta annotations? Move to an init method? Not use AOP? I don't know).
I am merely asking for a minimal setup that demonstrates the situation you were in with a single bean and the setup you had. This should also show what you mean by "not working": the bean lifecycle method is not called, or not in the right order, or things fail with AOP.

We can certainly help here but a minimal sample would save us a lot of back and forth and allow to focus on the actual issue.

Cheers

[cdprete]

I just opened this ticket to ask for clarifications from your side about some of claims done in that LinkedIn post. That's it.

I mentioned that migration from my side to confirm what @philwebb said and that's javax annotations are not supported anymore.
I'm not seeking here any help in that regard.

What I would like to understand here is if the claims done in that post around AOP and initialization order of @PostCustruct vs initMethod are true and, if it's so, how comes.

To me, initMethod looks like something that has been added in the past to make the Spring to Spring Boot migration easier for Spring developers.
So, in a Spring Boot project, I see no reason to use it instead of @PostConstruct.
On the other end, in that post there is a claim about @PostConstruct being called before the bean is fully initialized and that initMethod doesn't have this issue.

From my side, though, I fail to see how @PostConstruct may be called before the bean is fully initialized and that's why I'm asking in here for your opinions about it as experts in the subject.

[bclozel]

It is hard to argue about a LinkedIn post, especially if you are not the author nor reproduced the behavior claimed there.

As far as I know the javax annotations are still supported in Spring Framework 6.2.x (see https://github.com/spring-projects/spring-framework/blob/75b7f5efcf633abefe1d1e7ee58233facfc4ed9d/spring-context/src/main/java/org/springframework/context/annotation/AnnotationConfigUtils.java#L120 and spring-projects/spring-framework#27444).

I am not aware of any behavior not in line with the reference documentation that Phil pointed to already: postconstruct, init methods and bean lifecycle callback methods should be called after instantiation and property injection.

Maybe I am missing a specific case but I would need to see a concrete example of the behavior described there. Without that I don't think it's worth spending more time on this.

[cdprete]

Thanks for the details @bclozel.

That's actually interesting to see that javax.annotation should still be working when it was not the case for me, but that's not a big deal.

postconstruct, init methods and bean lifecycle callback methods should be called after instantiation and property injection.

That's what I was indeed expecting. :)

Maybe I am missing a specific case but I would need to see a concrete example of the behavior described there.

I would like to see one as well. 😅

It is hard to argue about a LinkedIn post, especially if you are not the author nor reproduced the behavior claimed there.

I agree, but my point here was about seeking some clarity.
I'm sorry if I gave you another impression.