> Well I agree with Semantic Versioning, so the question here that dictates > 3.2 vs 4.0 is whether we see Sonatype Aether as part of the exposed > supported API of Maven. IIRC the stated position is that plugin authors are > not supposed to rely on the Sonatype Aether API. If plugin authors have > relied on it then they are responsible for ensuring that the plugin works > in its absence... that would strongly indicate that 3.2 (or 3.1 if at the > same time as SLF4J) is the correct SemVer version... but if you view the > Sonatype Aether API as being part of the exposed supported API of Maven (as > opposed to leaked unsupported API) then 4.0 would be the correct SemVer
What if we published a specific policy that covered this case: something like: The Maven project is in transition in the area of dependency resolution. The existing official API is good enough for many problems, but does not expose enough operations to allow the correct implementation of a few, important plugins. Therefore, these plugins are coded directly to 'Sonatype Aether.' The Maven community plans to make changes to these APIs over the next few releases as we work to refine an appropriate public API in this area. Since these changes won't have much user-visible impact, they won't be major versions, even though they will change this API, and require the few plugins which touch it to change to track it. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org For additional commands, e-mail: dev-h...@maven.apache.org
