If getting local repo from settings.xml, do it right

[cstamas]

There are people who intentionally use non-default local repository locations, plus, they may have some expressions in the settings.xml...

So, if you do utilize code to get local repo from settings.xml, do it right. For me this whole stuff was not buildable due my (otherwise valid) settings.xml and non-default location of my local repository (that is on purpose not in default location).

[gnodet]

I would think the correct way to build settings is to use SettingsBuilder.

Though this is only used during tests, so that should be fine...

[cstamas]

Ideally, I'd drop all the deprecated stuff from maven-artifact as well (ArtifactRepo, layout...) as ultimately the goal here is "just" to get a file from local repo... To make it "proper" using SettingsBuilder and maybe SimpleLocalRepositoryManager from resolver, that would need all of maven and resolver (almost) just to perform this single thing...

Maybe the UT is wrong, and all we need is just to have this commons artifact under src/test/resources instead?

[kwin]

There is an easier solution in #357, please have a look.

[cstamas]

Did you try with settings.xml having ${env.XXX} inside? 😉

[slawekjaranowski]

IMHO tests are wrong, local repo is used to discover path for artifacts to build classpath ...
We need to check if we can simply give any path for assertions

[kwin]

Did you try with settings.xml having ${env.XXX} inside? 😉

Yes, because the SettingsBuilder does the interpolation (https://github.com/apache/maven/blob/42559937d8c4ed8a361422a8db3dfa6ef81fef21/maven-settings-builder/src/main/java/org/apache/maven/settings/building/DefaultSettingsBuilder.java#L235)

[slawekjaranowski]

suppressed by ##366 - now pass path to local repo from executing Maven