What does your SELECT statement for your cursor look like?

Were you also using OpenESQL and ODBC for accessing your Oracle database or were you using Pro*COBOL and COBSQL?

My select statement is : 

116000 EXEC SQL DECLARE CURPARMOD CURSOR FOR 
116100 SELECT 
116200 IDMOD, NOMBATCH, NOMTP 
116300 FROM 
116400 EG_PARMOD 
116800 END-EXEC. 
117000 EXEC SQL OPEN CURPARMOD END-EXEC. 

The problem is when i'm using Openesql and ODBC to access my Postgres database.

My script to compile my source file is :

cob -O -i -k my_file  -C SQL=DBMAN=ODBC -C SQL=BEHAVIOR=ANSI -C SQL=TARGETDB=POSTGRESQL -C SQL=ANSI92ENTRY -C SQL=CHECKSINGLETON -C SQL=CLOSE_ON_COMMIT=NO-C LIST=my_list_file -C CHKECM=CTRACE -C ECMLIST

My table in postgres database :

create table EG_PARMOD (
IDMOD CHAR(6) not null,
NOMBATCH CHAR(6) not null,
NOMTP CHAR(6) not null,
constraint PK_EG_PARMOD primary key (IDMOD) );

It looks like this internal pseudo column, ".oid", which is a unique identity column, is specific to PostgresSQL and was deprecated in V12 onward, but the OpenESQL run-time still requires this field in certain circumstances.

The oid column is required for “FOR UPDATE” cursor. The workaround is to create a real oid column in the table as documented here:
https://www.microfocus.com/documentation/enterprise-developer/ed90/ED-EclipseUNIX/GUID-1F1C4505-B771-4F8E-B274-952CAF3E8265.html

From the code snippet for the SELECT that you included, it doesn't look like you are using the “FOR UPDATE” cursor support., at least in this case so you might try adding the SQL(NOWHERECURRENT) directive to disable the use of oid in the runtime:

https://www.microfocus.com/documentation/enterprise-developer/ed90/ED-EclipseUNIX/GUID-DC52546D-EC5A-497B-A0CE-DB3BB882903B.html

If neither of these workarounds is suitable then please open up a support case and we can take a closer look. 

Thanks

Hi Chris,

SQL precompiler option : NOWHERECURRENT works fine !

Thanks a lot.

Regards,

Laurent.