Describe the bug
PR #7167 (fixes #7132) wrapped the document-shaped posting's write phase in a UnitOfWork and its own body names the sibling it left out: Posts.java.template (the flat per-item posts: mode) has the same multi-write shape with no unit - and its idempotency guard ("any row back-references this source") is coarser, so a tick that wrote SOME of its rows and failed is read as already-posted and the missing rows are never written. A half-post there is permanent, which is the same defect family as #7132/#7133 - named in the PR as out of scope, with no follow-up filed.
Expected
The per-item mode's writes for one source event run in one UnitOfWork, and/or the guard counts rows against the derived set instead of testing mere existence - a failed tick either leaves nothing or is completed by the redelivery.
Context
Named by PR #7167 (issue #7132). template-application-events-java/.../events/Posts.java.template.
Describe the bug
PR #7167 (fixes #7132) wrapped the document-shaped posting's write phase in a
UnitOfWorkand its own body names the sibling it left out:Posts.java.template(the flat per-itemposts:mode) has the same multi-write shape with no unit - and its idempotency guard ("any row back-references this source") is coarser, so a tick that wrote SOME of its rows and failed is read as already-posted and the missing rows are never written. A half-post there is permanent, which is the same defect family as #7132/#7133 - named in the PR as out of scope, with no follow-up filed.Expected
The per-item mode's writes for one source event run in one
UnitOfWork, and/or the guard counts rows against the derived set instead of testing mere existence - a failed tick either leaves nothing or is completed by the redelivery.Context
Named by PR #7167 (issue #7132).
template-application-events-java/.../events/Posts.java.template.