SEZ9 commented on issue #12601:
URL: https://github.com/apache/seatunnel/issues/12601#issuecomment-5987016492

   @SEPURI-SAI-KRISHNA thanks for following the namespace question through to 
`TypeRegistry` — this is the missing piece behind the earlier point that 
quoting alone cannot recover a type outside `search_path`.
   
   Your read matches what the previous review was asking for: the namespace is 
already reachable (the `pg_namespace` join is in `SQL_TYPES`), but it is 
discarded at the `addType` cache keying and again in `get(String name)` when 
the schema prefix is split off. So the "carry authoritative source type 
namespace to the sink DDL" requirement is a registry change plus plumbing to 
the sink, not a quoting-helper change. Please keep that work tied to #12606 
rather than a parallel fix, so there is one implementation path; if it is 
easier as a preparatory change, link it from #12606 and note the dependency 
there.
   
   Concrete asks on this point:
   
   1. Treat the three drop points you listed (select list, cache key, `get` 
lookup) as the scope of the namespace-carrying change, and make sure the 
end-to-end CDC-to-PostgreSQL schema-change test requested earlier covers the 
type-outside-`search_path` case through this path, with the type pre-created in 
the sink.
   2. Where the namespace still cannot be resolved, fail with a clear 
unsupported-type diagnostic rather than emitting an unqualified name.
   3. Since the multi-schema collision is read from the code and not yet 
reproduced, please verify it against a live Postgres before relying on it in 
the test design.
   
   On the `SQL_NAME_LOOKUP` / `loadType` first-row behaviour: yes, please open 
a separate issue for that, since it is independent of the DDL path and its fix 
(a namespace predicate and/or deterministic ordering like the sibling query) 
should be reviewed on its own. A short reproducer with two same-named types in 
different schemas would make that issue actionable.
   
   <!-- streview-comment:1520 -->


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to