I think that the way I would suggest this be solved is:
A) "define" where the R-margin is (probably after column 72 or after column
80 - each is used by some existing compilers)
B) If a LF (CRLF, whatever) appears OUTSIDE an alphanumeric literal, then
just treat it as the "end of source line".
C) If a LF (CRLF, whatever) appears "within" an alphanumeric literal, then
"logically" pad the line with enough spaces to the R-Margin.
D) There should be a compile-time (maybe run-time?) switch to allow ANSI
conforming behavior where a CR or LF *within* an alphanumeric literal is
treated as DATA. (This should be for "fixed form reference format" only).
***
Does this sound reasonable? Possible?
> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden On
> Behalf Of Sergey Kashyrin
> Sent: Thursday, January 19, 2006 7:24 PM
> To: Roger While
> Cc: address@hidden
> Subject: [open-cobol-list] Re: OC issues - number 4
>
> Roger,
>
> I've checked on OMVS subsystem on Mainframe (where the cobol
> source is not in a RECFM=F RECLEN=80)
> but the same format as on any unix (LF separated lines),
> and the test passed even there are no spaces at the end !
> It also passed on Microfocus and Fujitsu (not free format !)
>
> It seems to me that if "fixed" format for Cobol is used,
> compiler is padding with / assuming spaces up to 72-nd
> position anyway.
>
> So I suspect OC is wrong in this case.
> Once again, I'm noway a Cobol guru and don't know those
> details. Are there any formal descriptions of the rules ?
>
> BTW I might think to try to put OC on mainframe but currently
> libgmp is holding.
>
> Regards,
> Sergey
>
>
> ----- Original Message -----
> From: "Sergey Kashyrin" <address@hidden>
> To: "Roger While" <address@hidden>
> Cc: <address@hidden>
> Sent: Wednesday, January 18, 2006 3:23 PM
> Subject: [open-cobol-list] Re: OC issues - number 4
>
>
> > Roger,
> >
> > Yes, adding spaces helps.
> > Initially it's a mainframe programs and mainframe has a
> fixed record format, so original program has those spaces.
> > They've probably been stripped when transfering from mainframe.
> > But as a matter of fact both Fujitsu and Microfocus are
> accepting that.
> >
> > Regards,
> > Sergey.
> >
> > ----- Original Message -----
> > From: "Roger While" <address@hidden>
> > To: <address@hidden>
> > Cc: <address@hidden>
> > Sent: Wednesday, January 18, 2006 2:37 PM
> > Subject: Re: OC issues - number 4
> >
> >
> >> Let me guess - The line with the VALUE is immediately
> >> terminated after the apostrophe with a new-line
> >> and/or the following line is also terminated with a new-line
> >> immediately after the apostrophe.
> >> (At least BOTH are true when doing copy/paste from the post).
> >>
> >> This is then invalid.
> >> ie. The apostrophes would concatanate and be interpreted as
> >> one occurrence as per Cobol rules.
> >>
> >> Roger
> >>
> >>
> >> IDENTIFICATION DIVISION.
> >> PROGRAM-ID. ztest INITIAL.
> >> ENVIRONMENT DIVISION.
> >> DATA DIVISION.
> >> WORKING-STORAGE SECTION.
> >> 01 A.
> >> 10 FILLER PIC X(78) VALUE '
> >> - '
> >> - ' '.
> >> PROCEDURE DIVISION.
> >> DISPLAY 'A=' A ';'.
> >> GOBACK.
> >>
> >>
> >> Roger