Don't always trust Maven's '-f' arg value when resolving the current project pom.xml

[aloubyansky]

No description provided.

[geoand]

Tested against Quarkus platform and it works great

[famod]

Is this fixing strange Dev-Mode errors when using -f ./my-module instead of -f my-module?
I just had this yesterday and it took me a while to figure out what is going on...

[gsmet]

@aloubyansky there is a NPE in some of the tests:

io.quarkus.deployment.runnerjar.DependencyVersionOverridesManagedVersionTest
[ERROR] test  Time elapsed: 0 s  <<< ERROR!
java.lang.NullPointerException
	at java.nio.file.Files.provider(Files.java:97)
	at java.nio.file.Files.exists(Files.java:2385)
	at io.quarkus.bootstrap.resolver.maven.BootstrapMavenContext.getCurrentProjectPomOrNull(BootstrapMavenContext.java:626)
	at io.quarkus.bootstrap.resolver.maven.BootstrapMavenContext.getActiveSettingsProfiles(BootstrapMavenContext.java:462)
	at io.quarkus.bootstrap.resolver.maven.BootstrapMavenContext.resolveRemoteRepos(BootstrapMavenContext.java:391)
	at io.quarkus.bootstrap.resolver.maven.BootstrapMavenContext.getRemoteRepositories(BootstrapMavenContext.java:202)
	at io.quarkus.bootstrap.resolver.maven.MavenArtifactResolver.<init>(MavenArtifactResolver.java:118)
	at io.quarkus.bootstrap.resolver.maven.MavenArtifactResolver.<init>(MavenArtifactResolver.java:53)
	at io.quarkus.bootstrap.resolver.maven.MavenArtifactResolver$Builder.build(MavenArtifactResolver.java:89)
	at io.quarkus.bootstrap.resolver.ResolverSetupCleanup.initResolver(ResolverSetupCleanup.java:65)
	at io.quarkus.bootstrap.resolver.ResolverSetupCleanup.setup(ResolverSetupCleanup.java:30)
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)

[aloubyansky]

@famod will you be able to test whether it fixes your use-case? We may want to release 1.4.1.Final tomorrow including this fix.

[famod]

Is this PR actually trying to fix a regression in 1.4.0.Final?

I am asking because with 1.3.1.Final everything is ok when running mvn quarkus:dev -f my-module.
When I run mvn quarkus:dev -f ./my-module I get an AmbiguousResolutionException listing a class bean twice that exists in the very same module and is injected into a REST controller that also exists in the same module.
No difference with 1.3.2.Final.

But with 1.4.0.Final I also get this AmbiguousResolutionException when calling mvn quarkus:dev -f my-module, which works in 1.3.x.
The bean class is even listed three times when calling mvn quarkus:dev -f ./my-module.

And now comes the sad part: This PR does not seem to fix it. 😞

PS: Some gradlew download goes wrong in quarkus-platform-descriptor-json when building behind a proxy, although I have set all env vars etc.
For my test of this PR I removed the respective plugin from the pom.

[aloubyansky]

@famod is your app available somewhere on github to use as a reproducer?

[aloubyansky]

I think I know why @famod Will try to fix it today. Thanks for reporting.

[aloubyansky]

@famod just to make sure, it'll still be good if you could share a reproducer.

[aloubyansky]

@famod I pushed another commit here. Could you please give it a try?

[famod]

Could you please give it a try?

Will do, give me an hour.

it'll still be good if you could share a reproducer.

I'll try, but this will take me longer since I am not allowed to share the original project.

[aloubyansky]

Thanks!

[famod]

The new commit does not change anything for my case. :-/

I'll take a stab at a reproducer.

[aloubyansky]

Does it change anything if you use an absolute path with -f?

…

On Fri, Apr 24, 2020, 12:12 Falko Modler ***@***.***> wrote: The new commit does not change anything for my case. :-/ I'll take a stab at a reproducer. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#8794 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACO6M6EUNQHF2M3HCQ2NDDROFQZBANCNFSM4MPG7CTQ> .

[famod]

Update: I am still working on the reproducer but it seems this problem has to do with Maven CI friendly versioning (${revision} etc.)...

[aloubyansky]

@famod please open a new issue once you are ready. Thanks a lot!