Addressed comments.

CHANGELOG.md

@@ -26,6 +26,13 @@ Current
 
 ### Fixed:
 
+- [Fixes a bug where job metadata was being stored in the `ApiJobStore` even when the results came back synchronously](https://github.com/yahoo/fili/pull/49)
+    * The workflow that updates the job's metadata with `success` was running even when the query was synchronous. That update had the impact of
+      storing the ticket in the ApiJobStore.
+    * The reason the delay operator didn't stop the update workflow from executing viewed an `Observable::onCompleted` as a message for the purpose of
+      delays. Since the two observables that that the metadata update it gated on are empty when the query is synchronous, the update metadata workflow 
+      was being updated.
+    * The delay operator was replaced by `zipWith` as a gate.
 - [#45, removing sorting from weight check queries](https://github.com/yahoo/fili/pull/46)
 
 

fili-core/src/main/java/com/yahoo/bard/webservice/async/workflows/DefaultAsynchronousWorkflowsBuilder.java

@@ -176,6 +176,8 @@ private Observable<String> buildStorePreResponseChain(
         // We don't want to store the result in the PreResponseStore unless the query is asynchronous. The query is
         // asynchronous iff the asynchronousPayload emits at least one item.
         return preResponseEmitter
+                // Using zip as a gate. We don't let the preResponse from the preResponse emitter continue down the
+                // chain until and unless the asynchronousPayload emits an item.
                 .zipWith(asynchronousPayload, (preResponse, ignored) -> preResponse)
                 .flatMap(preResponse -> preResponseStore.save(jobRow.getId(), preResponse));
     }
@@ -206,7 +208,11 @@ private Observable<JobRow> buildUpdateRowChain(
         // The job status should not be updated until both the PreResponse has been stored, and the storage of the
         // original Job ticket has been attempted.
         return preResponseEmitter
+                // Using zip as a gate. We don't let the preResponse from the preResponse emitter continue down the
+                // chain until and unless the jobRowStoredNotification emits an item.
                 .zipWith(jobRowStoredNotification, (preResponse, ignored) -> preResponse)
+                // Using zip as a gate. We don't let the preResponse from the preResponse emitter continue down the
+                // chain until and unless the preResponseStoredNotification emits an item.
                 .zipWith(preResponseStoredNotification, (preResponse, ignored) -> preResponse)
                 .map(PreResponse::getResponseContext)
                 .map(responseContext -> responseContext.containsKey(ResponseContextKeys.ERROR_MESSAGE.getName()))