[
https://issues.apache.org/jira/browse/TRAFODION-1444?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14692672#comment-14692672
]
ASF GitHub Bot commented on TRAFODION-1444:
-------------------------------------------
Github user zellerh commented on a diff in the pull request:
https://github.com/apache/incubator-trafodion/pull/52#discussion_r36817313
--- Diff: core/sql/exp/ExpHbaseInterface.cpp ---
@@ -1315,32 +1192,30 @@ Lng32 ExpHbaseInterface_JNI::checkAndInsertRow(
const int64_t timestamp,
NABoolean asyncOperation)
{
- HTableClient_JNI *htc = client_->getHTableClient((NAHeap *)heap_,
tblName.val, useTRex_, hbs_);
- if (htc == NULL) {
- retCode_ = HBC_ERROR_GET_HTC_EXCEPTION;
- return HBASE_OPEN_ERROR;
- }
-
- Int64 transID;
+ HTableClient_JNI *htc;
+ Int64 transID;
+ NABoolean checkAndPut = TRUE;
+
if (noXn)
- transID = 0;
- else
+ transID = 0;
+ else
transID = getTransactionIDFromContext();
-
+ retCode_ = client_->insertRow((NAHeap *)heap_, tblName.val, hbs_,
--- End diff --
Since the method is called checkAndInsertRow, should we not call
checkAndInsertRow here as well? Sorry, haven't looked yet at the surround what
really happens, but this could either be wrong or confusing - unless I'm wrong
or confused :-)
> Reduce the path length for IUD operations in Trafodion in the hbase client
> layer
> --------------------------------------------------------------------------------
>
> Key: TRAFODION-1444
> URL: https://issues.apache.org/jira/browse/TRAFODION-1444
> Project: Apache Trafodion
> Issue Type: Improvement
> Components: sql-exe
> Affects Versions: 2.0-incubating
> Reporter: Selvaganesan Govindarajan
> Assignee: Selvaganesan Govindarajan
> Original Estimate: 96h
> Remaining Estimate: 96h
>
> There are 3 JNI to java transitions for insert, update, delete,
> checkAndInsert, checkAndUpdate and checkAndDelete operations. It can be done
> with one jni transition and reduced memory allocation.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
