Github user mjsax commented on a diff in the pull request:
https://github.com/apache/storm/pull/694#discussion_r37800211
--- Diff: storm-core/src/clj/backtype/storm/daemon/executor.clj ---
@@ -445,6 +445,32 @@
ret
))
+(defn init-batching-buffer [worker-context component-id]
+ (let [batchSize (.get_batch_size (.getComponentCommon worker-context
component-id))]
+ (if (> batchSize 1)
+ (let [consumer-ids (flatten (for [cids (vals (.getTargets
worker-context component-id))] (keys cids)))
+ consumer-task-ids (flatten (for [cid consumer-ids :when (not
(.startsWith cid "__"))] (into '() (.getComponentTasks worker-context cid))))]
+ (HashMap. (zipmap consumer-task-ids (repeatedly (count
consumer-task-ids) #(Batch. batchSize)))))
+ (HashMap.)
+ )))
+
+(defn emit-msg [out-task out-tuple overflow-buffer output-batch-buffer
transfer-fn]
+ (let [out-batch (.get output-batch-buffer out-task)]
+ (if out-batch
--- End diff --
Your observation is right for the current state of the code. However, I
would like to extend it with the possibility to used different batch sizes for
different output streams (including mixed-mode batching/non-batching).
Nevertheless, we could have 3 functions: non-batching, batching, mixed-mode.
Thus, we could choose the correct function at setup time. The branching
overhead is avoided for the both main cases (most times there is only a single
output stream).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
