[jira] [Commented] (NIFI-5515) [CaptureChangeMysql] Cannot get real value of primitive type when column type is unsigned.
[ https://issues.apache.org/jira/browse/NIFI-5515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17655140#comment-17655140 ] Matt Burgess commented on NIFI-5515: Turns out this will cause problems as NiFi Record types are all signed (so is Avro which we currently use for the schema). No solution would really be complete and I don't think we want to "half-support" unsigned values across NiFi. I looked at Debezium as well and they don't appear to support unsigned types either (except TINYINT(1) UNSIGNED which can be converted to a boolean). > [CaptureChangeMysql] Cannot get real value of primitive type when column type > is unsigned. > -- > > Key: NIFI-5515 > URL: https://issues.apache.org/jira/browse/NIFI-5515 > Project: Apache NiFi > Issue Type: Bug > Components: Extensions >Affects Versions: 1.6.0 >Reporter: ashwin konale >Priority: Minor > > *Processor: CaptureChangeMysql* > When mysql column type is of unsigned primitive type, CaptureChangeMysql is > not able to read the data properly when the value overflows. I investigated a > bit and came across this > ([https://github.com/shyiko/mysql-binlog-connector-java/issues/104|https://github.com/shyiko/mysql-binlog-connector-java/issues/104]). > The library which is internally being used in CaptureChangeMysql Processor > ([https://github.com/shyiko/mysql-binlog-connector-java|https://github.com/shyiko/mysql-binlog-connector-java]) > does not handle this case and its upto consumer to handle it. More > description can be found in the link. > *Issue summary Eg.* > Column type : unsigned tinyint > Actual value : 128 > Value from CaptureChangeMysql: -128 (Because its still 8bits storage and > mostsignificant bit is identified as sign from java.) > *Possible solution:* > Column types are read and stored Distributed Map Cache Client so we should be > able to do these conversions. > Nifi version : 1.6.0 > Mysql version : 5.7.22-log > Binlog type. : Row-based logging -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (NIFI-5515) [CaptureChangeMysql] Cannot get real value of primitive type when column type is unsigned.
[ https://issues.apache.org/jira/browse/NIFI-5515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17654083#comment-17654083 ] Matt Burgess commented on NIFI-5515: This information would have to be stored in the cache which would break compatibility. I recommend we ensure this capability works correctly in the CaptureChangeMySQLRecord processor under NIFI-4491. > [CaptureChangeMysql] Cannot get real value of primitive type when column type > is unsigned. > -- > > Key: NIFI-5515 > URL: https://issues.apache.org/jira/browse/NIFI-5515 > Project: Apache NiFi > Issue Type: Bug > Components: Extensions >Affects Versions: 1.6.0 >Reporter: ashwin konale >Priority: Minor > > *Processor: CaptureChangeMysql* > When mysql column type is of unsigned primitive type, CaptureChangeMysql is > not able to read the data properly when the value overflows. I investigated a > bit and came across this > ([https://github.com/shyiko/mysql-binlog-connector-java/issues/104|https://github.com/shyiko/mysql-binlog-connector-java/issues/104]). > The library which is internally being used in CaptureChangeMysql Processor > ([https://github.com/shyiko/mysql-binlog-connector-java|https://github.com/shyiko/mysql-binlog-connector-java]) > does not handle this case and its upto consumer to handle it. More > description can be found in the link. > *Issue summary Eg.* > Column type : unsigned tinyint > Actual value : 128 > Value from CaptureChangeMysql: -128 (Because its still 8bits storage and > mostsignificant bit is identified as sign from java.) > *Possible solution:* > Column types are read and stored Distributed Map Cache Client so we should be > able to do these conversions. > Nifi version : 1.6.0 > Mysql version : 5.7.22-log > Binlog type. : Row-based logging -- This message was sent by Atlassian Jira (v8.20.10#820010)
