PanacheRepository injected methods cannot be made transactional

[zeljkot]

Describe the bug
If a class implementing the PanacheRepository is marked as @Transactional, methods injected by Quarkus are not transactional.

Expected behavior
All public methods should be transactional. When calling e.g. persist, a transaction should be created and operation should succeed.

Actual behavior
App throws javax.persistence.TransactionRequiredException: Transaction is not active, consider adding @Transactional to your method to automatically activate one.

To Reproduce
Steps to reproduce the behavior:

1. Create a class implementing PanacheRepository and mark it as @Transactional
2. Create test class, with non-transactional test method and injected repository
3. Call persist from a non-transactional method.

Configuration

# Add your application.properties here, if applicable.

Screenshots
(If applicable, add screenshots to help explain your problem.)

Environment (please complete the following information):

• Output of uname -a or ver: Windows 10
• Output of java -version: 1.8
• GraalVM version (if different from Java):
• Quarkus version or git rev: 1.3.0.Alpha1

Additional context
(Add any other context about the problem here.)

[gsmet]

Yeah, I think it's probably the same issue of the methods being injected too late and not being taken into account by the interception layer.

/cc @mkouba

[agoncal]

Any update on this one ? I'm using 1.3.1.Final and I have the same issue (impossible to persist an entity with the repository because of transaction not being activated)

[mkouba]

@gsmet Yes, as described here the repository class is transformed and the methods are generated and so the CDI container does not see the methods at all.

[agoncal]

(cc @FroMage that's the issue I was mentioning during the podcast)

[FroMage]

@mkouba so, why do we run ArC before the Panache enhancers? How does ArC load the bytecode to see which methods need interceptors? ClassLoader, Jandex, or using an ASM transformer?

[mkouba]

so, why do we run ArC before the Panache enhancers?

@FroMage I have no idea when Panache enhancers are executed. By "run" you mean the analysis of the bean classes?

How does ArC load the bytecode to see which methods need interceptors?

We only consider methods from the Jandex.

[FroMage]

We only consider methods from the Jandex.

OK, so even if we ran ArC after we're done transforming, it would not see the new methods unless we reindexed them.

Is that the best thing to do, do you think @stuartwdouglas ?

[stuartwdouglas]

Isn't the underlying issue here just that ArC is not intercepting default methods? #9472

[mkouba]

@stuartwdouglas Nope, it's not about the default methods but about "synthetic" methods added during transformation... or at least that was my assumption ;-).

[FroMage]

OK, that's a very good point, because in this case all methods are defined on the superinterface. We're not introducing new methods, we're just overriding them.

But perhaps what @mkouba is saying is that even if the class itself is marked as @Transactional it does not apply to interface methods it inherits IFF the user hasn't overriden them? The fact that we override them later via a transformer makes it too late for ArC to consider them overridden.

[mkouba]

We're not introducing new methods, we're just overriding them.

Aha, that's a completely different situation then. I'm sorry for confusion. In that case, Stuart's PR should work fine.

[FroMage]

OK great. Note that I've filed #9487 for interception of static methods.