https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126541
Bug ID: 126541
Summary: cobol1: error recovery: implicit define missing data
items
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: cobol
Assignee: unassigned at gcc dot gnu.org
Reporter: simonsobisch at gnu dot org
Target Milestone: ---
in case when there's a typo in the DATA DIVISION - or a missing/wrong copybook,
the same variable is error'd all over the place - and quickly fills
-fmax-errors
test.cob:658:23: error: DATA-ITEM ‘S-GEWM’ not found
658 | COMPUTE S-GEWM = S-GEWSALDOZSM /
| ^~~~~~~~~
test.cob:661:25: error: DATA-ITEM ‘Z-S-GEWM’ not found
661 | MOVE 0 TO S-GEWM.
| ^~~~~~~~~
test.cob:689:25: error: DATA-ITEM ‘Z-S-GEWM’ not found
689 | MOVE 0 TO S-GEWM.
| ^~~~~~~~~
test.cob:702:17: error: DATA-ITEM ‘S-GEWM’ not found
702 | MOVE S-GEWM TO DNL-GEWM.
| ^~~~~~~~~
test.cob:777:25: error: DATA-ITEM ‘Z-S-GEWM’ not found
777 | MOVE 0 TO S-GEWM.
| ^~~~~~~~~
test.cob:956:25: error: DATA-ITEM ‘Z-S-GEWM’ not found
956 | MOVE 0 TO S-GEWM.
| ^~~~~~~~~
I suggest to internally define it, preferably per its use; which is what
GnuCOBOL does.
Example: the COMPUTE above would copy the S-GEWSALDOZSM defintion in - or
simply defines a PIC S9(18)v(12) item.
If there is a MOVE "ABCD" TO UNDEF-VAR it would be defined as PIC X(4), or
bigger...
That would be similar for other user-defined names like "OPEN fd-name".
Also note the cc1 way of "not defined"... and then suppress all follow-up
warnings on that same data-item.
