[
https://issues.apache.org/jira/browse/SLING-12464?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17893971#comment-17893971
]
Konrad Windszus commented on SLING-12464:
-----------------------------------------
I currently don't see a use case for this superset/subset logic in regions.
IMHO each region should independently of another region specify its own
exported packages. There is no need for inheritance AFAICS except for the fact
that the global region(s) are implicitly visible to all other ones.
So although I see a use case for a tree of depth 2 like this (where A is the
global region)
{code}
A
/ \
B C
{code}
I fail to see a use case for a deeper tree like
{code}
A
/ \
B C
/
D
{code}
where D is a subset of B.
> Allow disjunct API regions
> --------------------------
>
> Key: SLING-12464
> URL: https://issues.apache.org/jira/browse/SLING-12464
> Project: Sling
> Issue Type: Improvement
> Components: Feature Model
> Affects Versions: Feature Model API Regions Extension 2.0.0
> Reporter: Konrad Windszus
> Priority: Major
>
> Currently API regions always form a line, i.e. one is always a superset of
> the other. The documentation at
> https://github.com/apache/sling-org-apache-sling-feature-extension-apiregions/blob/master/docs/api-regions.md#visibility-of-java-api
> explicitly states:
> bq. Exports declared earlier in the api-regions array also apply to later
> elements in the array
> That currently prevents having completely disjunct API regions (not sharing
> any exported-packages).
> A use case for disjunct API regions was discussed in
> https://lists.apache.org/thread/mh36p586vqhxsvkjz6lsgmtf1cz3d99g.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
