Bearing in mind that the registered RFC server programs are programmed differently in UNICODE/NON-UNICODE mode, the Unicode test for registered RFC server programs can cause the RFC server program to crash, or it can trigger an exception in the server program. The user of transaction SM59 receives the I message "The character size cannot be determined."
Solution
If Unicode is set in transaction SM59 under the "Special options" tab as the character width in the target system, and the registered RFC server program is not Unicode-enabled, the Unicode test in SM59 can cause the registered RFC server program to crash. In this case, the RfcLastErrorEx call should always follow after a RfcAcceptEx call fails.
If non-Unicode is set in transaction SM59 under the "Special options" tab as the character width in the target system, and the registered RFC server program is Unicode-enabled, the Unicode test can trigger different responses in SM59 depending on the RFC server program:
1. A .NET server reacts with an ABAP exception and triggers the exception:
Exception:
FUNCTION NOT IMPLEMENTED
Please note that the.NET server has not crashed and that it is still logged onto the SAP gateway.
2. A JCo server reacts with an RFC abort (see trace):
"JCo.Server could not find server function '???? ' "
In this case, a message appears, stating
"Check the unicode ability of the server program"
Note that the JCo server has not crashed and that it is therefore still registered on the SAP gateway.
Note:
You can establish whether the RFC server program is Unicode-enabled as follows:
On Unix:
what <executable-name> | grep RFC
If librfcu* is found, the program is linked with the Unicode library.
On NT:
find "LIBRFC" "<executable-name>"
If librfc32U.dll is displayed, the program is linked with the UNICODE library. If librfc32.dll is displayed, the program is linked with the non-Unicode library.
On Java:
JCo is always Unicode-enabled as of Version 2.0. In this case, the non-Unicode variant of the librfc is also used internally in the Unicode case.
Whether or not a JCo server runs in Unicode mode depends on whether or not the property
"jco.server.unicode" was set to 1 when the JCo server was created.
(1=Unicode mode; 0=Non Unicode mode [default])
However, for communication with Unicode systems, it is absolutely necessary that Unicode mode is used and that the character width in the target system is set to Unicode in the target system in SM59. Non-Unicode communication with Unicode systems is not supported by the JCo. This type of function would not make sense because Java is only ever based on a Unicode character set.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.