A WebSphere Transformation Extender map in a WebSphere Enterprise Service Bus mediation flow corrupts or truncates text output on Linux platforms

Cause

Java Virtual Machines (JVMs) use different encoding on different platforms. For example, the JVM on a Microsoft® Windows® system uses code page 1252 (LATIN 1) encoding, while the JVM on a Linux system uses UTF-8 encoding.

In a mediation flow, WebSphere ESB can alter the data depending on the default JVM encoding of the platform. For example, on a Linux system, WebSphere ESB can convert the data to UTF-8 before the data reaches WebSphere Transformation Extender, which changes the output data from a WebSphere Transformation Extender map.

When the WebSphere Transformation Extender map primitive receives a business object that is an instance of string data, such as comma-separated value (CSV) data, and the map primitive uses a type tree (.MTT file) instead of a native schema, the map primitive receives the data in the native-encoding of the JVM on that platform.

When the type tree is defined as native schema, on output the type tree converts the input data to UTF-8 encoding to match the schema.

• On a Windows system, the type tree correctly converts the LATIN 1-encoded data to UTF-8 encoding; for example, by converting 0xDF to 0xC39F.
• On a Linux system, the type tree converts data that is already UTF-8 encoded; for example, by converting 0xC39F to 0xC383C29F. The conversion causes the data corruption or truncation.