[
https://issues.apache.org/jira/browse/TRAFODION-1453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14717055#comment-14717055
]
ASF GitHub Bot commented on TRAFODION-1453:
-------------------------------------------
Github user sureshsubbiah commented on a diff in the pull request:
https://github.com/apache/incubator-trafodion/pull/67#discussion_r38118768
--- Diff: core/sql/executor/ExHdfsScan.cpp ---
@@ -1427,24 +1424,28 @@ char *
ExHdfsScanTcb::extractAndTransformAsciiSourceToSqlRow(int &err,
else
attr = NULL;
- if (!isTrailingMissingColumn)
- sourceColEnd = strchr(sourceData, delimiter);
-
- if(!isTrailingMissingColumn)
- {
- short len = 0;
- if (sourceColEnd &&
- sourceColEnd <= sourceDataEnd)
- len = sourceColEnd - sourceData;
- else
- {
- len = sourceDataEnd - sourceData;
- if (i != hdfsScanTdb().convertSkipListSize_ - 1)
- isTrailingMissingColumn = TRUE;
- }
-
- if (attr) // this is a needed column. We need to convert
- {
+ if (!isTrailingMissingColumn) {
+ sourceColEnd = hdfs_strchr(sourceData, rd, cd, sourceDataEnd,
&rdSeen);
--- End diff --
What about NULL? We used to treat \N as NULL before I think. Can we still
do that. For non-string columns will a missing value be treated as NULL?
> A null character in hive TEXTFILE format is not reported as an error
> --------------------------------------------------------------------
>
> Key: TRAFODION-1453
> URL: https://issues.apache.org/jira/browse/TRAFODION-1453
> Project: Apache Trafodion
> Issue Type: Bug
> Components: sql-exe
> Affects Versions: 0.6 (pre-incubation)
> Reporter: Selvaganesan Govindarajan
> Fix For: 2.0-incubating
>
>
> Trafodion engine expects the data in hive TEXTFILE format shouldn't have any
> null character. Currently, Trafodion supports only ISO88591 and UTF8
> character set string data only. When there is a null characterthe engine
> continues to read all the data without producing any output.
> An error should be reported instead.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
