appclient container needs 2.1.1 Jakarta Annotation ManagedBean

[scottmarlow]

Signed-off-by: Scott Marlow smarlow@redhat.com

Possible change for #24014

[scottmarlow]

Note that this doesn't build correctly for me yet due to Could not find artifact org.jboss.weld.se:weld-se-shaded:jar:5.0.SP2

[dmatej]

Oh, nice, so I can drop my local changes where I "hacked" appclient by adding newer annotation-api to the classpath before weld-se-shaded.

[arjantijms]

@scottmarlow If I look here, there's no 5.0.SP2 yet:

https://repo1.maven.org/maven2/org/jboss/weld/se/weld-se-shaded

The last one is

5.0.0.SP2         2022-06-08 13:08         -

[starksm64]

That artifact is there:
https://repo1.maven.org/maven2/org/jboss/weld/se/weld-se-shaded/5.0.0.SP2/

weld-se-shaded-5.0.0.SP2-javadoc.jar              2022-06-08 13:08   7205180      
weld-se-shaded-5.0.0.SP2-javadoc.jar.asc          2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2-javadoc.jar.md5          2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2-javadoc.jar.sha1         2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2-sources.jar              2022-06-08 13:08   2913339      
weld-se-shaded-5.0.0.SP2-sources.jar.asc          2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2-sources.jar.md5          2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2-sources.jar.sha1         2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2.jar                      2022-06-08 13:08   3464828      
weld-se-shaded-5.0.0.SP2.jar.asc                  2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2.jar.md5                  2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2.jar.sha1                 2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2.pom                      2022-06-08 13:08      7598      
weld-se-shaded-5.0.0.SP2.pom.asc                  2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2.pom.md5                  2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2.pom.sha1                 2022-06-08 13:08        40 

[starksm64]

There will have to be a new post 5.0.0.SP2 release to pickup the update of common annotations to 2.1.1.

[dmatej]

That artifact is there: https://repo1.maven.org/maven2/org/jboss/weld/se/weld-se-shaded/5.0.0.SP2/

weld-se-shaded-5.0.0.SP2-javadoc.jar              2022-06-08 13:08   7205180      
weld-se-shaded-5.0.0.SP2-javadoc.jar.asc          2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2-javadoc.jar.md5          2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2-javadoc.jar.sha1         2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2-sources.jar              2022-06-08 13:08   2913339      
weld-se-shaded-5.0.0.SP2-sources.jar.asc          2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2-sources.jar.md5          2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2-sources.jar.sha1         2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2.jar                      2022-06-08 13:08   3464828      
weld-se-shaded-5.0.0.SP2.jar.asc                  2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2.jar.md5                  2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2.jar.sha1                 2022-06-08 13:08        40      
weld-se-shaded-5.0.0.SP2.pom                      2022-06-08 13:08      7598      
weld-se-shaded-5.0.0.SP2.pom.asc                  2022-06-08 13:08       659      
weld-se-shaded-5.0.0.SP2.pom.md5                  2022-06-08 13:08        32      
weld-se-shaded-5.0.0.SP2.pom.sha1                 2022-06-08 13:08        40 

This one was already used, see dates, 8th june. The version 5.0.SP2 instead of 5.0.0.SP2 seems suspicious to me ...

[dmatej]

Seems also there are no other changes in Weld core github repo, including weld-se-shaded project. So we should probably create a PR with updated dependencies for them.

[dmatej]

Aha, I get it ... we need to update weld-core's dependency on weld-api ...

[dmatej]

See weld/core#2747

[scottmarlow]

weld/core#2747 + https://issues.redhat.com/browse/WELD-2722 sound like a great change but considering that the pull request is still in draft mode and the work in my opinion, sounds like something that might not be solved quickly, I wonder if it is better to hack in the individual common annotations 2.1.1 jar into the appclient container as @dmatej was likely doing before I created this broken pr! :-)

[dmatej]

My "hack" wasn't nice and @manovotn estimates that tomorrow they will release new Weld, so we can probably wait for that :-)

[scottmarlow]

I changed the pull request to update to Weld 5.0.0.SP3 (looks like next likely tag in https://github.com/weld/core/tags), if that is not the right version (when released), I will update this pr again.

[manovotn]

@scottmarlow The version will be 5.0.1.Final (using the Weld API 5.0.SP2).
The release is running now so if there are no issues with it, I assume it could land in Central within an hour or so.

[dmatej]

@scottmarlow The version will be 5.0.1.Final (using the Weld API 5.0.SP2). The release is running now so if there are no issues with it, I assume it could land in Central within an hour or so.

Just FYI, it reached Maven Central, co I initiated rebuild on Jenkins.

[dmatej]

Eh, Maven default rules - it saves failures, so it will recheck after 24 h again. Another option is to purge the Maven repository.

[dmatej]

Seems OSGi doesn't like this change ...

[arjantijms]

After a quick run it seems like the CDI TCK also doesn't like this upgrade:

[ERROR] Failures: 
[ERROR]   ProcessBeanAttributesNotFiredForBuiltinBean>Arquillian.run:138->testProcessBeanAttributesNotFired:51 expected [true] but found [false]
[ERROR]   AfterTypeDiscoveryTest>Arquillian.run:138->testFinalAlternatives:128 expected [true] but found [false]
[ERROR]   AfterTypeDiscoveryTest>Arquillian.run:138->testInitialAlternatives:81 expected [3] but found [11]
[ERROR]   AfterTypeDiscoveryMassOperationsTest>Arquillian.run:138->testInitialAlternatives:77 expected [3] but found [11]

Maybe a Weld 5.0.1 is just too much for now? What we really need is a 5.0.0.SP3, which only changes the problematic Jakarta Annotation ManagedBean and absolutely nothing else. No comment, no whitespace, nothing. Just ManagedBean.

[dmatej]

After a quick run it seems like the CDI TCK also doesn't like this upgrade:

[ERROR] Failures: 
[ERROR]   ProcessBeanAttributesNotFiredForBuiltinBean>Arquillian.run:138->testProcessBeanAttributesNotFired:51 expected [true] but found [false]
[ERROR]   AfterTypeDiscoveryTest>Arquillian.run:138->testFinalAlternatives:128 expected [true] but found [false]
[ERROR]   AfterTypeDiscoveryTest>Arquillian.run:138->testInitialAlternatives:81 expected [3] but found [11]
[ERROR]   AfterTypeDiscoveryMassOperationsTest>Arquillian.run:138->testInitialAlternatives:77 expected [3] but found [11]

Maybe a Weld 5.0.1 is just too much for now? What we really need is a 5.0.0.SP3, which only changes the problematic Jakarta Annotation ManagedBean and absolutely nothing else. No comment, no whitespace, nothing. Just ManagedBean.

Finally we found that these CDI TCK failures were caused by something between Krazo and CDI TCK, it has nothing to do with this PR.

But with this PR GlassFish simply doesn't start, because Weld requires jboss-logging >=3.5.0