Unique dataset creation

hariibm · New User Joined: 05 Feb 2007 Posts: 61 Location: Chennai

Hi,

I am coding a REXX utility/macro and it creates an output file.

What would be the best approach to ensure that it creates the unique dataset every time this utility is run.

Regards,
Harikrishna.

hariibm · New User Joined: 05 Feb 2007 Posts: 61 Location: Chennai

Please note this utility can be run multiple times in a day.

dbzTHEdinosauer · Posted: Tue Oct 12, 2010 2:01 pm

your ALLOCAT command will return an error when you attempt to allocate a file as new, when it already exists.
so code to re-enter you name generating algorithm.

but then again, I assume you meant 'unique dataset name'.
If not, I apologize for dropping by this thread.

kacks7 · Posted: Tue Oct 12, 2010 2:05 pm

hi hariibm,
use the date and time combination for the qualifiers of your dataset.

hariibm · New User Joined: 05 Feb 2007 Posts: 61 Location: Chennai

Hello Dick Brenholtz,
Yes, exactly. My macro is failing when I am trying to run second time.
I am creating the dataset like below:
userid.mon+days.xxx
ie, If userid = YAF8, Mon = Oct, days = 285, then the dataset would be YAF8.OCT285.XXX

I want this utility to create a different dataset next time when it is run and it should not give any error. so please let me know how to create a unique dataset.

It is ok if the existing dataset can be used. But the ALLOCAT command try to create new. please guide how to go ahead.

Regards,
Hari

notonly4u · New User Joined: 26 Apr 2005 Posts: 87 Location: Hyderabad

Can't GDGs be used?

dbzTHEdinosauer · Posted: Tue Oct 12, 2010 2:30 pm

one second you say unique - where Kacks7's suggestion is appropriate
and the next you say that you can reuse existing DS, which means you can use a disposition of MOD.

what actually is the use of the DS? and don't say output of the REXX.

USERID as HLQ isolates the DS to each user.
if there is a requirement for multiple DS to exist, date & time are easy ways to name a file.
If there is no need to maintain multiple DS for each user, make it a standard name (fixed name), and use MOD.

or you could do what many 3rd party tools do:

create a temporary dataset,
invoke a Browse or VIEW
and leave it to the user to make a copy if there is a requirement to save the file.

notonly4U,
though GDG's could be used,
a gdg base would be required for every user

hariibm · New User Joined: 05 Feb 2007 Posts: 61 Location: Chennai

No. GDGs can't be used.

expat · Posted: Tue Oct 12, 2010 2:57 pm

Any particular reason why a GDS is inappropriate ?

dbzTHEdinosauer · Posted: Tue Oct 12, 2010 3:56 pm

one of the things I enjoy about this forum is that it does reflect a microcosm of the world.

87 questions have been asked of the TS,
and what does he provide as a response?.

superk · Posted: Tue Oct 12, 2010 4:52 pm

Well, if I had to do this, and I wanted to assure unique dataset names, I'd consider using some or all of the following within the nodes:

- Current Date (MMDDYY or YYYYDDD or Base Date).
- Current Time (HHMMSS or # of seconds since midnight).
- System ID if it might run on different LPAR's.
- If the process was running as a batch job, I also might include the jobname, jobid, account.

dick scherrer · Posted: Wed Oct 13, 2010 2:05 am

Hello,

If one becomes really really creative when generating a unique dataset name it may be very difficult to reference the dataset later. . .

Kurt Deininger · Posted: Wed Oct 13, 2010 3:00 pm

Hi Hariibm,

A simple solution would be to create a file having a counter being updated each time(set it to zero after creation):

expat · Posted: Wed Oct 13, 2010 4:12 pm

Again I ask ................................