On Sun, 19 Jun 2016 17:25:45 +0100, Rupert Reynolds wrote:
>I can see an obvious need to avoid corruption of the directory. Yet there
>is (or was) valid code (using BLKSIZE=256 from memory) to access a PDS
>directly. And wasn't IEBGENER with a dummy SYSUT1 used to repair the
>DSORG=PO of a PDS damaged in this way? So I think it would be difficult to
>make sensible changes to the access methods.
>
The user should be allowed any desired overrides to RECFM, LRECL,
BLKSIZE, and DSORG opening a data set for INPUT. The data set
remains intact. Access methods or DFSMS should prohibit opening
for OUTPUT or UPDATE that would modify existing attributes.
If such rules had originally been in force, there would have been no
damage, and no need to repair.
>I'd suggest making sure there are checks in all the IBM utilities, as in
>the RFE, as a practical answer.
>
I consider it less practical to make changes in *ALL* IBM utilities, past,
present, and future, than in one central place in DFSMS.
Suggestion to OP. Use:
//STEP EXEC PGM=IEBGENER
//SYSPRINT DD SYSOUT=(,)
//SYSIN DD *
//SYSUT2 DD SYSOUT=(,)
//SYSUT1 DD DISP=SHR,DSN=data.set.name(member1),
// RECFM=FB,LRECL=80,BLKSIZE=32720
... to recover any member.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
