Index _all_ dependencies with a certain groupId

[famod]

Description

For the migration to Quarkus 3, I'm about to remove jandex-maven-plugin from my ~20 modules Maven project to have one thing less to keep in sync for each Quarkus update.

As I don't really like empty beans.xml and listing all artifactIds is pretty verbose (especially in a distributed config setup), having Quarkus index all artifacts that belong to a configured groupId would really help.
Relevant docs: https://quarkus.io/guides/cdi-reference#how-to-generate-a-jandex-index

Implementation ideas

It probably boils down to making quarkus.index-dependency.<name>.artifact-id optional.

[quarkus-bot]

/cc @Ladicek (arc), @manovotn (arc), @mkouba (arc)

[famod]

Also /cc @geoand

[manovotn]

As I don't really like empty beans.xml

Is there some actual reason why this wouldn't work for you or is it just a matter of preference?
Adding a marker file seems like, by far, the easiest solution here that wouldn't require any further work even if using distributed config setup. Or is there something I am missing?

[famod]

Empty beans.xml do absolutely work. It's 20 empty files vs. one single config setting though.

[famod]

PS: external libs are a different story w.r.t. beans.xml.

[manovotn]

PS: external libs are a different story w.r.t. beans.xml.

Sure this one I can understand and that's what the configs are primarily for.

Empty beans.xml do absolutely work. It's 20 empty files vs. one single config setting though.

Right, but do they hurt anyhow? You basically add it and forget about it, not need for config as well.
It sure is possible to add what you're asking for, I am just curious why are marker files so bad.

Also note that CDI 4 uses this syntax to mark a bean archive in CDI Lite, see https://jakarta.ee/specifications/cdi/4.0/jakarta-cdi-spec-4.0.html#bean_archive

[famod]

Right, but do they hurt anyhow? You basically add it and forget about it, not need for config as well.

They don't, it's just more inconvenient. You have to remember to add them to new modules, for example (rare case, admittedly).
I'm not saying empty beans.xml files are bad per se. It's just about developer joy and getting things done quickly and neatly and DRY. 😉

[geoand]

I'm +0 on this

[Ladicek]

This could be done using an extension I think? Collect all the relevant vartifacts from the CurateOutcomeBuildItem and produce IndexDependencyBuildItem for them. Then, you don't even have to configure anything :-)

[gsmet]

I think it could be a valuable addition if you're willing to do the work.

[mkouba]

If we decide to implement this then it could be more practical to turn the artifact-id into a regex (or support wildcards?) so that it's possible to select a subset of artifacts with a given groupId.