Reading SPR-2744 it seems like the expected behaviour is to inherit scope from a parent bean - that makes sense to me, but it doesn't happen in practice. So maybe this is a regression?

Indeed, inner beans inherit the scope from their containing beans. I'll check whether that actually still works in the 2.1 codebase (if you have a specific test case around, please attach it!). I guess we should also consider validating inner bean definitions more aggressively, rejecting scopes that do not match the parent's.

Juergen

Attached a test case.

There are actually two issues here - inner beans do not inherit their parent's scope, and inner beans cannot be scoped explicitly (as per the issue description). I would like to see both fixed if that's possible. Shall we open another issue?

Attached another versio nof test cases that explicitly test both scenarios

Inner beans can now be scoped explicitly. Also, when no scope is provided for an inner-bean it will inherit the containing bean's scope by default (even for 'custom' scopes). Both of the attached test-cases pass.