hubcio commented on code in PR #4739:
URL: https://github.com/apache/iggy/pull/4739#discussion_r4474504870


##########
core/connectors/sinks/iceberg_sink/src/router/dynamic_router.rs:
##########
@@ -119,61 +80,46 @@ impl Router for DynamicRouter {
                 continue;
             };
 
-            let message = match writer.push_to_existing(&route_field_val, 
message) {
-                Some(msg) => msg,
-                None => continue,
+            let entry = match tables.entry(route_field_val) {
+                Entry::Occupied(mut entry) => {
+                    entry.get_mut().1.push(message);
+                    continue;
+                }
+                Entry::Vacant(entry) => entry,
             };
+            let route_field_val = entry.key();
 
-            if !is_valid_namespaced_table(&route_field_val) {
+            if !is_valid_namespaced_table(route_field_val) {
                 warn!(
                     "Dropping the message at offset {}: its route field value: 
{route_field_val} names no table. It needs at least one namespace before the 
table, separated by '.'",
                     message.offset
                 );
                 continue;
             }
 
-            let route_field_val_cloned = route_field_val.clone();
-
-            if writer
-                .load_table_if_exists(&route_field_val_cloned, 
self.catalog.as_ref())
-                .await
-                .is_ok()
-            {
-                if let Some(msgs) = 
writer.table_to_message.get_mut(&route_field_val_cloned) {
-                    msgs.push(message);
-                } else {
-                    let message_vec: Vec<ConsumedMessage> = vec![message];
-                    writer
-                        .table_to_message
-                        .insert(route_field_val_cloned, message_vec);
-                }
-            } else {
+            let Some(table) = table_exists(route_field_val, 
self.catalog.as_ref()).await else {
                 warn!(
                     "Dropping the message at offset {}: table: 
{route_field_val} does not load from the catalog",
                     message.offset
                 );
-            }
-        }
-
-        for (table_name, table_obj) in &writer.tables_to_write {
-            let batch_messages = match 
writer.table_to_message.remove(table_name) {
-                Some(m) => m,
-                None => continue,
+                continue;
             };
+            entry.insert((table, vec![message]));
+        }
 
-            let data: Vec<Payload> = batch_messages.into_iter().map(|m| 
m.payload).collect();
-
-            write_data(
-                &data,
-                table_obj,
+        for (table_name, (table, batch_messages)) in &tables {

Review Comment:
   warning: `DynamicRouter` keeps payloads for completed tables alive until 
every table commits, increasing memory use during later uploads. consume 
`tables` by value and borrow each table and message vector for `write_data`.



-- 
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