Avoid recomputing new_pipeline_ids to increase readability. (#3215)

quickwit/quickwit-indexing/src/actors/indexing_service.rs

@@ -472,14 +472,18 @@ impl IndexingService {
 
  let running_pipeline_ids: HashSet<IndexingPipelineId> =
  self.indexing_pipeline_handles.keys().cloned().collect();
- let new_pipeline_ids = updated_pipeline_ids
+
+ let new_pipeline_ids: Vec<&IndexingPipelineId> = updated_pipeline_ids
  .difference(&running_pipeline_ids)
- .collect_vec();
+ .collect();
 
+ // We fetch the new indexes metadata.
  let indexes_metadata_futures = new_pipeline_ids
  .iter()
+ // No need to emit two request for the same `index_id`
  .unique_by(|pipeline_id| pipeline_id.index_id.clone())
  .map(|pipeline_id| self.index_metadata(ctx, &pipeline_id.index_id));
+
  let indexes_metadata = try_join_all(indexes_metadata_futures).await?;
  let indexes_metadata_by_index_id: HashMap<String, IndexMetadata> = indexes_metadata
  .into_iter()
@@ -489,7 +493,7 @@ impl IndexingService {
  let mut failed_spawning_pipeline_ids: Vec<IndexingPipelineId> = Vec::new();
 
  // Add new pipelines.
- for new_pipeline_id in updated_pipeline_ids.difference(&running_pipeline_ids) {
+ for new_pipeline_id in new_pipeline_ids {
  info!(pipeline_id=?new_pipeline_id, "Spawning indexing pipeline.");
  let index_metadata = indexes_metadata_by_index_id
  .get(&new_pipeline_id.index_id)