Konrad Windszus
Konrad Windszus
> I don't expect it to be responsible for execution decisions. The loading with extensions has no impact on goal execution.
For me the decision whether a plugin should be loaded with or without extensions should primarily be driven by the plugin developer (not by the consumer). In general plugins usually...
> It seems to me that this was a technical decision rather than a design one. For me the underlying design principle is that all plugins not being loaded with...
> I think packaging and artifact handlers (aka dependency types) are 2 separate topics This is fine, but doesn't have any impact on this PR, does it?
@hboutemy Do you still have concerns with merging this? If so can you please elaborate on the why?
Also look at https://github.com/apache/maven-shared-utils/pull/54. IMHO we should only implement in one library....
@slawekjaranowski This depends on #109 being merged first otherwise the build breaks even before executing the unit test. Update: Done and rebased this PR.
Yes, exactly. In this case the target does not (yet) exist. The question is how `FileUtils.copyFile(...)` should behave in that case.
IMHO copy should also not fail in case the source symlink is broken (which it currently does)
> IMHO copy should also not fail in case the source symlink is broken (which it currently does) I added another (failing) test case for this scenario in https://github.com/apache/maven-shared-utils/pull/110/commits/8ff14a538e13399fbc307c6c3d8cf98437104364