HHH-15283 - fix NPE for NamedNativeQuery + SqlResultSetMapping (columns)

[NathanQingyangXu]

https://hibernate.atlassian.net/browse/HHH-15283

This bug was from https://hibernate.atlassian.net/browse/HHH-15070, a suite of bug reports submitted by Spring Data JPA team member for their Hibernate v6 support (see spring-projects/spring-data-jpa#2423 (comment))

This PR aim to solve one confirmed bug; there might be other PRs for the same ticket.

Seems another NPE bug. In the following entity:

	@SqlResultSetMapping(
		name = "mapping",
		columns = @ColumnResult( name = "cnt" )
	)
	@NamedNativeQuery(
		name = "sample.count",
		resultSetMapping = "mapping",
		query = "SELECT count(*) AS cnt FROM Sample"
	)
	@Entity(name = "Sample")
	static class Sample {

		@Id
		@GeneratedValue
		Long id;

	}

At least two classes' NPE issues are exposed.

[dreab8]

Hi @NathanQingyangXu ,

it seems to me that https://hibernate.atlassian.net/browse/HHH-15070 targets only https://github.com/odrotbohm/hibernate-bugs/tree/main/h6-embedded-id-record Custom EmbeddedInstantiator not considered for embedded identifier types so I think it is better if you create a specific Jira for this PR.

Thanks

[NathanQingyangXu]

Hi @NathanQingyangXu ,

it seems to me that https://hibernate.atlassian.net/browse/HHH-15070 targets only https://github.com/odrotbohm/hibernate-bugs/tree/main/h6-embedded-id-record Custom EmbeddedInstantiator not considered for embedded identifier types so I think it is better if you create a specific Jira for this PR.

Thanks

Yeah, I tested all the bugs under https://github.com/odrotbohm/hibernate-bugs, but I agree we'd better create a new ticket for better tracking.

[hibernate-github-bot]

Thanks for your pull request!

This pull request appears to follow the contribution rules.

› This message was automatically generated.

[NathanQingyangXu]

I created a new ticket here: https://hibernate.atlassian.net/browse/HHH-15283 and explained its relationship with https://hibernate.atlassian.net/browse/HHH-15070. The git commit message prefix was also changed.

[dreab8]

Thanks @NathanQingyangXu

[odrotbohm]

Thanks, folks, for spending time with this. As Nathan correctly spotted, the repo lined in the ticket indeed contains reproduces for quite a few issues we found. That said, each of the folders in that repo already has a ticket filed. We usually point to the folder within the repository for each ticket. A quick search for odrotbohm/hibernate-bugs reveals this.

[NathanQingyangXu]

Thanks, folks, for spending time with this. As Nathan correctly spotted, the repo lined in the ticket indeed contains reproduces for quite a few issues we found. That said, each of the folders in that repo already has a ticket filed. We usually point to the folder within the repository for each ticket. A quick search for odrotbohm/hibernate-bugs reveals this.

Yeah, this link points to the new ticket I just created.

[dreab8]

I'm going to close https://hibernate.atlassian.net/browse/HHH-15070, as explained by @NathanQingyangXu it is fixed in 6.0.1.Final

[odrotbohm]

That's nice, thanks Andrea! I'm going to take a round through the reproducer repo and document the latest state of them in readmes of each of the modules.

[dreab8]

Thnaks @odrotbohm

[dreab8]

@NathanQingyangXu I have made a little change to this PR #5064, what do you think?

[odrotbohm]

I'm going to close https://hibernate.atlassian.net/browse/HHH-15070, as explained by @NathanQingyangXu it is fixed in 6.0.1.Final

Unfortunately, the issue is not gone. I've updated the reproducer, and it still fails, not applying the custom EmbeddableInstantiator.

[dreab8]

I think the issue is the use of record for the UserId

[NathanQingyangXu]

Good catch! Thanks.

…

On Thu., May 19, 2022, 6:08 a.m. Andrea Boriero, ***@***.***> wrote: @NathanQingyangXu <https://github.com/NathanQingyangXu> I have made a little change to this PR #5064 <#5064>, what do you think? — Reply to this email directly, view it on GitHub <#5055 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB6UYASOTMWU5TCSESM7TNTVKYHIDANCNFSM5V5BZSBA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[odrotbohm]

Which I create and register a custom EmbeddableInstantiator for, which is not used. That is precisely what the ticket is about. UserId is an embeddable. EmbeddableInstantiators can be registered to customize their instantiation. Still, it is not used when used as id. It is used when UserId is used as non-id property (the test gets green if you try).

That also shows that the fact that the class is a record, is not the culprit. In fact, records not working as embeddables, out of the box is the driver behind this ticket.

(I apologize for hijacking this ticket, but I unfortunately cannot comment on tickets in your JIRA.)

[NathanQingyangXu]

I will take a new look later and try to fix it if record is not the reason. Generally speaking, record is incompatible with JPA stuff IMO.

…

On Thu., May 19, 2022, 7:16 a.m. Oliver Drotbohm, ***@***.***> wrote: Which I create and register a custom EmbeddableInstantiator for, which is not used. That is precisely what the ticket is about. UserId is an embeddable. EmbeddableInstantiators can be registered to customize their instantiation. Still, it is not used when used *as id*. It *is* used when UserId is used as non-id property (the test gets green if you try). That also shows that the fact that the class is a record, is not the culprit. In fact, records not working as embeddables, out of the box is the driver behind this ticket <https://hibernate.atlassian.net/browse/HHH-15072>. (I apologize for hijacking this ticket, but I unfortunately cannot comment on tickets in your JIRA.) — Reply to this email directly, view it on GitHub <#5055 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB6UYAUS5CJWI5YP5KHJ6I3VKYPI5ANCNFSM5V5BZSBA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[dreab8]

I tried and yes it happend that if you do not use the record the test is green but it is true that the EmbeddableInstantiator is not called, so I'm going to re-open the Jira.

Thanks @odrotbohm and @NathanQingyangXu

[dreab8]

Thanks @NathanQingyangXu ,

I'm going to close this in favour of #5064