Host-variable in File Section : DSNH310I W with Db2 preprocessor (passing) vs IGYPS0231-S with integrated coprocessor (blocking)

Hi,

We study the transition from the Db2 preprocessor to the Db2 coprocessor integrated into the COBOL compiler.

We come across cases of programs that use hosts-variables which are file record areas.

The Db2 preprocessor simply emits a warning and the compilation proceeds (afterwards the developer is responsible for any errors at runtime): DSNH310I W

DSNH310I W     DSNHSMUD LINE 408 COL 28  COBOL HOST VARIABLE "E12L-DCJYDT" WAS DECLARED IN FILE SECTION    
DSNH310I W     DSNHSMUD LINE 408 COL 45  COBOL HOST VARIABLE "E12L-DCJYFI" WAS DECLARED IN FILE SECTION   
DSNH310I W     DSNHSMUD LINE 409 COL 42  COBOL HOST VARIABLE "E12L-PEGL00" WAS DECLARED IN FILE SECTION 

The Db2 coprocessor blocks compilation by issuing a severe level message: IGYPS0231-S        

   430  IGYPS0231-S   SQL host variable "E12L-DCJYDT" was defined in the "FILE SECTION".  The statement was discarded.
   430  IGYPS0231-S   SQL host variable "E12L-DCJYFI" was defined in the "FILE SECTION".  The statement was discarded.
   431  IGYPS0231-S   SQL host variable "E12L-PEGL00" was defined in the "FILE SECTION".  The statement was discarded.
   601  IGYPS0226-E   DSNH504I DSNHSMUD LINE 601 COL 22  CURSOR "JW-CURSOR" WAS NOT DECLARED

We can already see a different behavior about severity level between the preprocessor and the Db2 coprocessor.

But is it fundamentally justified to make a severe error?

With SQLDA management induced by the coprocessor, the memory addresses of the host-variables are recalculated before each access, so a priori no risk of error?

Unlike the operation of the preprocessor which declares as many SQLDAs as SQL accesses and which by default only calculates the addresses of the hosts-variables once during the first SQL access.

The severe error level therefore does not seem justified, and a warning level as the preprocessor did would seem sufficient.

It remains possible to modify the severity level of the IGYPS0231 message, by setting up an MSGEXIT, but it would seem more appropriate that it be a warning level, identical to the preprocessor, and leave the possibility of worsening the level by an MSGEXIT.