Hello there fellow COBOLics,

we found something, that looks like a defect in the COBOL 6.3 Compiler. We have a variable length table that is defined like

01  TEMP-OUT.        03 ODO-Object              PIC S9(4) BINARY.    03 ODO-Subject    OCCURS 0 TO 42 DEPENDING ON ODO-Object.        05 Tabular-Data-1       PIC X(69).        05 Tabular-Data-2       PIC X(69).        05 Tabular-Data-n       PIC X(69).

and then the SORT looks like:

SORT ODO-Object   ASCENDING Tabular-Data-1            Tabular-Data-2

Now, whenever ODO-Object contains ZERO, Message IGZ0279W is issued for each data-name given in the SORT Statement as a result of the NUMCHECK option. Since sorting an empty table is a rather trivial task, we would not expect those fields to be touched at all. 
Language Reference in regard to SORT states pretty clearly: "The number of occurrences of table elements that are referenced by data-name-2 is determined by the rules in the OCCURS clause." And the Programming Guide in regard to ODO states "The length of the variable portions of a group item is the product of the object of the DEPENDINGON phrase and the length of the subject of the OCCURS clause."

Can someone confirm this behaviour. Is there a reason for this behaviour or is it really a defect?

Kind regards,
Christoph

Hello there fellow COBOLics,

we found something, that looks like a defect in the COBOL 6.3 Compiler. We have a variable length table that is defined like

01  TEMP-OUT.        03 ODO-Object              PIC S9(4) BINARY.    03 ODO-Subject    OCCURS 0 TO 42 DEPENDING ON ODO-Object.        05 Tabular-Data-1       PIC X(69).        05 Tabular-Data-2       PIC X(69).        05 Tabular-Data-n       PIC X(69).

and then the SORT looks like:

SORT ODO-Object   ASCENDING Tabular-Data-1            Tabular-Data-2

Now, whenever ODO-Object contains ZERO, Message IGZ0279W is issued for each data-name given in the SORT Statement as a result of the NUMCHECK option. Since sorting an empty table is a rather trivial task, we would not expect those fields to be touched at all. 
Language Reference in regard to SORT states pretty clearly: "The number of occurrences of table elements that are referenced by data-name-2 is determined by the rules in the OCCURS clause." And the Programming Guide in regard to ODO states "The length of the variable portions of a group item is the product of the object of the DEPENDINGON phrase and the length of the subject of the OCCURS clause."

Can someone confirm this behaviour. Is there a reason for this behaviour or is it really a defect?

Kind regards,
Christoph

Hello there fellow COBOLics,

we found something, that looks like a defect in the COBOL 6.3 Compiler. We have a variable length table that is defined like

01  TEMP-OUT.        03 ODO-Object              PIC S9(4) BINARY.    03 ODO-Subject    OCCURS 0 TO 42 DEPENDING ON ODO-Object.        05 Tabular-Data-1       PIC X(69).        05 Tabular-Data-2       PIC X(69).        05 Tabular-Data-n       PIC X(69).

and then the SORT looks like:

SORT ODO-Object   ASCENDING Tabular-Data-1            Tabular-Data-2

Now, whenever ODO-Object contains ZERO, Message IGZ0279W is issued for each data-name given in the SORT Statement as a result of the NUMCHECK option. Since sorting an empty table is a rather trivial task, we would not expect those fields to be touched at all. 
Language Reference in regard to SORT states pretty clearly: "The number of occurrences of table elements that are referenced by data-name-2 is determined by the rules in the OCCURS clause." And the Programming Guide in regard to ODO states "The length of the variable portions of a group item is the product of the object of the DEPENDINGON phrase and the length of the subject of the OCCURS clause."

Can someone confirm this behaviour. Is there a reason for this behaviour or is it really a defect?

Kind regards,
Christoph

Hello there fellow COBOLics,

we found something, that looks like a defect in the COBOL 6.3 Compiler. We have a variable length table that is defined like

01  TEMP-OUT.        03 ODO-Object              PIC S9(4) BINARY.    03 ODO-Subject    OCCURS 0 TO 42 DEPENDING ON ODO-Object.        05 Tabular-Data-1       PIC X(69).        05 Tabular-Data-2       PIC X(69).        05 Tabular-Data-n       PIC X(69).

and then the SORT looks like:

SORT ODO-Object   ASCENDING Tabular-Data-1            Tabular-Data-2

Now, whenever ODO-Object contains ZERO, Message IGZ0279W is issued for each data-name given in the SORT Statement as a result of the NUMCHECK option. Since sorting an empty table is a rather trivial task, we would not expect those fields to be touched at all. 
Language Reference in regard to SORT states pretty clearly: "The number of occurrences of table elements that are referenced by data-name-2 is determined by the rules in the OCCURS clause." And the Programming Guide in regard to ODO states "The length of the variable portions of a group item is the product of the object of the DEPENDINGON phrase and the length of the subject of the OCCURS clause."

Can someone confirm this behaviour. Is there a reason for this behaviour or is it really a defect?

Kind regards,
Christoph