WIP: allow OCCURS DEPENDING ON with REDEFINES clauses
#259
Conversation
lefessan
added
the
MF-AIX
GitMensch
left a comment
GitMensch
left a comment
There was a problem hiding this comment.
I've took 20 minutes to check the dialect support.
Please adjust the comments/valuas as noted and add cobc/Changelog and config/Changelog to your files..
Please also add runtime tests that show what you see on MF - and that GC with the new option has the same result.
| record-contains-depending-clause: obsolete | ||
| defaultbyte: none # "not specifically defined in Standard COBOL" | ||
| picture-l: unconformable | ||
| odo-in-redefines: unconformable # to be checked |
There was a problem hiding this comment.
xopen is only stricter than cobol85+amendment, so that's unconformable
| record-contains-depending-clause: obsolete | ||
| defaultbyte: 0 | ||
| picture-l: ok | ||
| odo-in-redefines: unconformable # to be checked |
| record-contains-depending-clause: unconformable | ||
| defaultbyte: 0 # not verified yet, but likely to be as IBM | ||
| picture-l: unconformable | ||
| odo-in-redefines: unconformable # to be checked |
| record-contains-depending-clause: unconformable | ||
| defaultbyte: " " | ||
| picture-l: unconformable | ||
| odo-in-redefines: warning |
There was a problem hiding this comment.
the warning is only raised when compiled with "cautions enabled" - so it should be "ok" here (and a possible cobc_acu wrapper will take the necessary -a parameter and translate it which will include -fodo-in-redefines=warning)
... but for larger-redefines it should actually be warning - please adjust that when changing this file
| record-contains-depending-clause: unconformable | ||
| defaultbyte: 0 # not verified, but likely like IBM | ||
| picture-l: unconformable | ||
| odo-in-redefines: unconformable # to be checked |
There was a problem hiding this comment.
Haven't found anything about it in its COBOL Reference Guide [that's in windows HLP format...] it has some rules on OCCURS but doesn't specify anything between that and REDEFINES.
In any case that would be # not verified, but likely like IBM
3 participants
As of today this PR only allows the parsing of those cases, the codegen is to be checked