RfcSvr and Unicode

[Former Member]

Hi, gurus!

My question is: does DI RfcSvr supports unicode? When I trying to test unicode RFC connection to RfcSvr it falls to core dump (on Solaris 10 SPARC)...

[scott_broadway]

Hi Sergey,

Can you tell us a little more about your environment? For instance:

-four-part version of Data Integrator/Data Services

-SAP version, including whether the system is set to use unicode

-Locale on the Solaris environment that runs your jobserver (output from "locale" command)

Many thanks,

~Scott

[Former Member]

We have a DI job what transfers data from Oracle Database to SAP BW info-source. The job works perfectly but where are small problem with cyrillic character encoding - cyrilic letters shown at BW as non-cyrilic symbols. Codepages settings for entire dataflow are following:

Oracle Database codepage - CL8MSWIN1251

NLS_LANG setting on Job Server host - RUSSIAN_CIS.CL8MSWIN1251

Job Server OS - Slaris 10 Generic_137111-05 SPARC

Job Server account locale - en_US.UTF-8

SAP BW is unicode system

Codepage for Oracle DS in DI - cp1251

Codepage for SAP BW - we have tried cp1251 and utf-8

In DI Designer we can wiev Oracle data correctly. If we transfer data to file on Job Server host rather SAP BW DS - we got correct data in all cases (cp1251 and utf-8) at file format. But not in SAP BW.

For BW in case of utf-8 we got two-character sequences for each character from Oracle (because DI RfcSvr doesn't support unicode - isn't it?).

In case of cp1251 we got symbols with following character codes for russian alphabet:

For capital letters: 49152, 49408, 49664, 49920, 50176, etc...

For lowercase letters: 57344, 57600, 57856, 58112, 58368, etc...

Now we use self-developed FM at BW side to convert whis symbols to correct russian symbols - but it's not rught method...

Thanks...

Edited by: Sergey Ivanov on Oct 20, 2008 11:11 AM

Edited by: Sergey Ivanov on Oct 20, 2008 11:12 AM

[werner_daehn]

The rfcsvr.exe is not involved in the data movement. It just gets a signal "start batch job now with this RequestID" and that's it.

I would have thought that you did everything correct, especially setting the BW datastore to utf-8 as codepage.

The only other thought I would have is that the stty settings of Unix have an impact on the SAPGUI. "stty cs8 -istrip"

Just a guess...

[Former Member]

I've solve my problem...

It seems that SAP BW does not completly supports unicode - russian characters transfers by RFC as single-byte caharcters with conversion. So i've set environment variable SAP_CODEPAGE=1504 and librfc became transfer correct characters.

[werner_daehn]

Thanks, didn't know that!

[Former Member]

Hi,

I have the same problem but in a Windows server. I haven´t found information about how to set the code_page to receive the RFC invocation from Unicode SAP Server.

Could you please help me?

[Former Member]

We have the same problem.

We've made an integration between SAP BW and external sources using

data services. So the problem is that the system uploads incorrect data

with cyrillic symbols. We cannot see russian letters, the system shows

incomprehensible symbols.

E.g.: we upload a table, which includes columns with data: 1. name

(type "char") 2. lastname (type "char") 3. postal codes

(type "integer"). Column №3 uploaded succesfully but columns №2 and №1

were uploaded with incorrect symbols. Data in English upload

succesfully.

GUI Incoding for import/export data uses ANSI for Unicode systems.

SAP BusinessObjects Data Services locale selector uses codepage UTF-8.

BW targe codepage cp1251

Also we added system variable "SAP_CODEPAGE" with value 1504 to

settings in Windows 7.

We try to change this settings, but hasn't right effect.

Need your help to solve this problem.

[Former Member]

We solve this problem after change Codepage for SQL DS in DI - cp1251