Re: Table Definitions DDS or SQL?

Joep Beckeringh


rob@xxxxxxxxx wrote:

Like Denis, most of our files are defined using DDS. But that's packaged software for you.

I define my new files using SQL.  Then again, I am probably the only one.

Here's something to chew over:

Problem with DDS created files. DDS created files are built to allow crap in most fields. Some legacy thing when people got off grunting, thumping their chest and writing batch programs that checked fields down to the byte level for data that is right. Like the first COBOL program I wrote in a class - make sure a field that is supposed to be numeric, is. DDS files are fast to write to, but slow to read from because they don't care what kind of data you write to them. But they'll check the data quality on a read. When you think about it, most of your data access is reading, right?

SQL created files check data on a write.  Not on a read.
Recently saw a classic example of this.
Compile this:
                R JOEPR
                  MYCHAR         5A

MYPACK 9P 2 with CRTPF FILE(ROB/JOEP) SRCFILE(ROB/QDDSSRC)

Run this command:  CRTPF FILE(ROB/JOEP2) RCDLEN(10)
STRSQL
INSERT INTO ROB/JOEP2 VALUES('NOWISTHETI')

CREATE TABLE ROB/JOEP3 (MYCHAR CHAR (5 ) NOT NULL WITH DEFAULT, MYPACK DEC (9 , 2) NOT NULL WITH DEFAULT)

RCDFMT JOEP3R
(exit sql)

CPYF FROMFILE(JOEP2) TOFILE(JOEP) MBROPT(*ADD) FMTOPT(*NOCHK)

Notice: Allowed crap into JOEP. You have a packed decimal field with alpha characters in it.

But CPYF FROMFILE(JOEP2) TOFILE(JOEP3) MBROPT(*ADD) FMTOPT(*NOCHK)
Tells me to bugger off with CPF5035-Data mapping error on member JOEP3.

Now if I attempt to read the corrupted DDS created file with a simple RPGLE cycle program

     FJOEP      IP   E             DISK
     FQSYSPRT   O    F   80        PRINTER
     OQSYSPRT   D
     O                       MYCHAR
     O                                              ' '

O MYPACK I will get RNQ0907-Decimal-data error occurred.

Verifying that DDS files check data integrity on read. SQL created files check data integrity on write.

Rob Berendt