Released Clover.ETL 2.4

21. 2. 2008

New version (2.4) of Clover.ETL engine has been released. It brings many improvements in graph execution, database communication and also extensions of Clover Transformation Language (CTL)

Most notably, we have introduced new graph execution management which allows us to run several transformation graphs within the same JVM instance at the same time. Also, it is possible to limit the amount of CPU resources (threads) used by Clover during graph's runtime. Other interesting extension is ability to run transformations in transaction mode - transformation graph can be re-set to its original state - that allows processing of data in small chunks - similarly to database "triggers".

List of other (smaller) improvements:

• sftp protocol support for readers which use multifile reader
• possibility of copying ByteDataField as BLOB
• new implementation of "partition by range" feature with range lookup table usage
• field unquoting before setting value by delimited and universal data parsers
• the set of available functions in the Clover transformation language was extended by more than 20 new functions
• Clover transformation language supports two new variable types, byte array and record
• some components have new optional "rejected" port(s) (AproxMergeJoin, CheckForeignKey, DBJoin, LookupJoin)
• DBOutputTable accepts more queries at once; new jetel-sql query syntax introduced
• DBOutputTable has new attribute "errorAction" - COMMIT or ROLLBACK on errors
• StructureWriter has 3 input ports: "body", "header" and "footer". There can be different masks for each input port
• all component attributes, which contain a java code or a Clover TL language code, are not resolved for special characters
• new graph processing facility - GraphExecutor, GraphPool
• new component XmlWriter with example graphXmlWriter.grf
• new component SpeedLimiter
• XLSWriter can write to different sheets depending on record's value
• data record metadata can specify a default field delimiter
• the DataReader component recognizes too few or too many fields per record
• the DataReader component has extended error port metadata - field number where an error occurs is attached
• new JavaExecute component
• new component attribute type - properties
• ascending/descending flag in sort component can be specified per each key field