I have a strange issue which I will probably report as a bug but thought i
would pass by you guys 1st.

I have an update statement

EXEC SQL UPDATE HARDVER SET
HAR_IMEPRT=NVL(RTRIM(:HAR-IMEPRT),' ')
,HAR_IMETER=NVL(RTRIM(:HAR-IMETER),' ')
,HAR_WINDOW=NVL(RTRIM(:HAR-WINDOW),' ')
,HAR_TIPTER=NVL(RTRIM(:HAR-TIPTER),' ')
,HAR_TIPPRT=NVL(RTRIM(:HAR-TIPPRT),' ')
,HAR_CEKPRT=NVL(RTRIM(:HAR-CEKPRT),' ')
,HAR_MJESTO=NVL(RTRIM(:HAR-MJESTO),' ')
,HAR_TELBRO=NVL(RTRIM(:HAR-TELBRO),' ')
,HAR_PORUKA=NVL(RTRIM(:HAR-PORUKA),' ')
,HAR_POSJED=:HAR-POSJED
,HAR_KESBOY=:HAR-KESBOY
,HAR_AMBROJ=:HAR-AMBROJ
,HAR_PMBROJ=:HAR-PMBROJ
,HAR_STATUS=:HAR-STATUS
,HAR_REDBRO=:HAR-REDBRO
,HAR_UPDATE=:HAR-UPDATE
,HAR_SATMIN=:HAR-SATMIN
,HAR_TELLER=:HAR-TELLER
,HAR_CEKOVI=:HAR-CEKOVI
,HAR_PRTOFF=:HAR-PRTOFF
WHERE ROWID=:HARDVER-ROWID
END-EXEC

All of the host variables have values in them.

For some reason HAR_PMBROJ is being updated with a null even thougth
HAR-PMBROJ contains 84.

If I comment out one of the assignments above HAR_PMBROJ then HAR_PMBROJ is
ok but the next column in the update statement HAR_STATUS gets a null. If I
comment out 2 of the assignments above HAR_PMBROJ then HAR_REDBRO gets a
null.

This is very strange and I have no idea why it is happening.

TIA

--
Steve

Op 8-1-2010 22:04, Steve Rainbird schreef:
What happens if you put NVL's around all the assignments? (Or at least
the first 4 or 5?)
This can determine whether the error is in the update or in reading the
host variables.

Shakespeare

On Jan 8, 4:04*pm, "Steve Rainbird"
<steve.nospam.rainb... (AT) mssint (DOT) nospam.com> wrote:
What is the total length of all of these variables? Have you exceeded
the line length restriction in COBOL or for Pro*COBOL? Since this
NULL 'walks' the variable list as you comment out preceeding entries
it might be a bug that's reading the line terminator as a value.

All pure speculation, to be honest.


David Fitzjarrell

"Shakespeare" <whatsin (AT) xs4all (DOT) nl> wrote

If I put NVL around all the numeric fields then HAR_PMBROJ gets updated with
zero (HAR-PMBROJ contains 84).

--
Steve

"ddf" <oratune (AT) msn (DOT) com> wrote

This is the definition of the variables.

07 HAR-IMEPRT PIC X(20).
07 HAR-IMETER PIC X(20).
07 HAR-WINDOW PIC X(20).
07 HAR-TIPTER PIC X(10).
07 HAR-TIPPRT PIC X(10).
07 HAR-CEKPRT PIC X(20).
07 HAR-MJESTO PIC X(15).
07 HAR-TELBRO PIC X(10).
07 HAR-PORUKA PIC X(6).
07 HAR-POSJED PIC 9(3).
07 HAR-KESBOY PIC 9(6).
07 HAR-AMBROJ PIC 9(2).
07 HAR-PMBROJ PIC 9(2).
07 HAR-STATUS PIC 9(1).
07 HAR-REDBRO PIC 9(8).
07 HAR-UPDATE PIC 9(8).
07 HAR-SATMIN PIC 9(8).
07 HAR-TELLER PIC 9(1).
07 HAR-CEKOVI PIC 9(1).
07 HAR-PRTOFF PIC 9(1).

Steve

I

--
Steve

On Jan 8, 5:33*pm, "Steve Rainbird"
<steve.nospam.rainb... (AT) mssint (DOT) nospam.com> wrote:
If I remember my COBOL, Pic 9(2) is numeric character, that is, the
character format of a numeric digit: 0 - 9. Is the raw values signed
or over-punched by any chance. If so PIC S9(02) should fix the
issue.

Does your code use null indicator bytes? Are these set correctly. I
have never coded pro*COBOL but in pro*C you have to set the length
bytes in the corresponding Oracle structure for variable character
columns. If this needs to be done in COBOL were the corresponding
length bytes for the first few character fields set correctly.

HTH -- Mark D Powell --

"Mark D Powell" <Mark.Powell2 (AT) hp (DOT) com> wrote

Thanks,

No the digits are not overpunched in any way. but I will try it with signed
fields and see if that makes any difference.

There no indicator variables used, and in any case if they were they would
have to be specified in the UPDATE statement.

:HAR-PMBROJ:HAR-PMBROJ-IND or something like that I seem to remember.

--
Steve

Op 8-1-2010 22:54, Shakespeare schreef:
From this one must conclude that the variables are not read anymore.
The update part is working. Seems some buffer is oveflowed.

Shakespeare