[
https://issues.apache.org/jira/browse/KYLIN-4757?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17212418#comment-17212418
]
ASF subversion and git services commented on KYLIN-4757:
--------------------------------------------------------
Commit ec980712f93f9fdde639a6957e500897073bb163 in kylin's branch
refs/heads/master-hadoop3 from Xiaoxiang Yu
[ https://gitbox.apache.org/repos/asf?p=kylin.git;h=ec98071 ]
KYLIN-4757 Fix decimal precision
> Impossible precision for decimal datatype in kylin if the source column is
> numeric in postgres
> ----------------------------------------------------------------------------------------------
>
> Key: KYLIN-4757
> URL: https://issues.apache.org/jira/browse/KYLIN-4757
> Project: Kylin
> Issue Type: Bug
> Components: RDBMS Source
> Affects Versions: v3.1.0
> Reporter: André Luís da Rosa de Lima
> Assignee: Xiaoxiang Yu
> Priority: Major
> Labels: postgresql, source-jdbc
> Fix For: v3.1.1
>
>
> Using Postgres as data source, columns of datatype numeric without a
> explicitly set precision and scale become "decimal(256)" in kylin. 256 comes
> from the value of "kylin.source.hive.default-varchar-precision" (despite
> using jdbc as datasource). This causes an error when building a cube.
> The problem can be fixed in
> "org.apache.kylin.source.jdbc.extensible.JdbcExplorer" class
> ([https://github.com/apache/kylin/tree/
> master/source-jdbc/src/main/java/org/apache/kylin/source/jdbc/extensible/JdbcExplorer.java#L111|https://github.com/apache/kylin/blob/master/source-jdbc/src/main/java/org/apache/kylin/source/jdbc/extensible/JdbcExplorer.java#L111]
>
> [)|https://github.com/apache/kylin/blob/kylin-4.0.0-alpha/source-jdbc/src/main/java/org/apache/kylin/source/jdbc/extensible/JdbcExplorer.java#L111)]
> The csize variable defined in line 96 receives the value 131089 from
> columns.getInt("COLUMN_SIZE");
> I fixed it by modifying the following part in the source file starting after
> line 111:
> {code:java}
> int precision = (SqlUtil.isPrecisionApplicable(kylinType) && csize > 0) ?
> csize : -1;
> precision = Math.min(precision,
> KylinConfig.getInstanceFromEnv().getDefaultVarcharPrecision());
>
> int scale = (SqlUtil.isScaleApplicable(kylinType) && digits > 0) ? digits :
> -1;
> {code}
> to this:
>
> {code:java}
> int precision = (SqlUtil.isPrecisionApplicable(kylinType) && csize > 0) ?
> csize : -1;
> int maxPrecision = precision;
> if (kylinType.equals("char")) {
> maxPrecision = KylinConfig.getInstanceFromEnv().getDefaultCharPrecision();
> } else if (kylinType.equals("varchar")) {
> maxPrecision =
> KylinConfig.getInstanceFromEnv().getDefaultVarcharPrecision();
> } else if ((kylinType.equals("decimal") || kylinType.equals("numeric"))) {
> maxPrecision =
> KylinConfig.getInstanceFromEnv().getDefaultDecimalPrecision();
> }
> precision = Math.min(precision, maxPrecision);
> int scale = (SqlUtil.isScaleApplicable(kylinType) && digits > 0) ? digits :
> -1;
> {code}
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
