Special characters in result parameters are not displayed correctly when returned to the calling Cobol batch program from zRES rule execution
Some special NLS characters (eastern european languages, for example Z WITH CARON) are not displayed correctly when returned to the calling Cobol batch program from zRES rule execution. All is well when tested from RD with UTF-8 code page, and also when using SOAP-UI client.
Results from Java rules engine include the correct character (Unicode Character 'LATIN SMALL LETTER Z WITH CARON' U+017E). Java (Unicode) uses 16 bits and therefore has support for a very large number of characters.
Now the client application needs to handle and display this character glyph. The client application is responsible for using a codepage which includes all the necessary characters and glyphs.
In Rule Designer or SoapUI, you use UTF-8 which can handle vast numbers of glyphs by using multiple bytes when necessary. With an 8-bit EBCDIC/ASCII codepage however, you can only display 256 characters. So the problem is finding an 8-bit codepage which includes all these worldwide NLS characters. This is in fact impossible; there are only regional 8-bit codepages.
The Cobol mainframe program can receive a UTF-8 byte array, i.e. an array of bytes which should be interpreted as UTF-8, and move it around. A 3270 terminal cannot correctly display the glyph. If the COBOL gave the byte-array (character string) to a web browser - then it would probably appear correctly.
If the consuming application is not able to display this special character, you might replace it with a valid English character.
To summarize:
The problem is not with ODM, but with the code page of client application (specified as Copybook import code page) and probably the 3270 terminal