Have you tried...

_n_

easy example:

data yourstep (keep= whatever this that here there); set example.sasdata; if _n_ > 500 then stop; /* you'll get 499, it will read the 500th ob(record) then stop */ run;

or note the difference......

data yourstep (keep= whatever this that here there); set example.sasdata; if your logic here; if _n_ > 500 then stop; /* you'll get 499, it will read the 500th ob(record) then stop */ run; in this last example because you have logic (if , where, /* whatever */ ) it does not mean you'll get 499!

the _n_ is just a counter of where sas is in the PDV execution loop. its very handy for testing code prior to letting it loose on MONSTER datasets and debugging long/intricate jobs. you could build bigger conditions with _n_.

hth. =)

zt.

>---------- >From: Jeff J. Voeller[SMTP:atrocity@MY-DEJANEWS.COM] >Sent: Wednesday, December 23, 1998 2:53 PM >To: SAS-L@UGA.CC.UGA.EDU >Subject: OBS= equivalent for output datasets? > >We all know that SAS has that nifty little OBS= option for running tests on >small portions of input datasets. But is there a way to apply similar logic >to an output dataset? > >In other words, I want to process a DATA step until the result data reaches a >certain number of observations, regardless of how many observations in the >input dataset need to be read. > >For those familiar with SyncSort, what I'm suggesting is akin to how its >STOPAFT= parameter works. > >I guess I could initialize a counter and then STOP when it hits my desired >number... > >Thanks for any suggestions. > > >-----== Sent via Deja News, The Discussion Network ==----- >http://www.dejanews.com/ Easy access to 50,000+ discussion forums >