Date:         Sat, 27 Jun 1998 18:23:37 GMT
Reply-To:     Ron Stanonik <stanonik@NPRDC.NAVY.MIL>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Ron Stanonik <stanonik@NPRDC.NAVY.MIL>
Organization: Navy Personnel R&D Center
Subject:      subsetting large raw file?

Sorry if this is novice question.  I've searched the
various FAQ's.
 
We're trying to create a subset of a large raw file (1.5GB).
The large file, history, is sorted by an ID number.
If we wanted a subset of only a few ID numbers, we'd do
something like
 
        data subset;
        infile history;
        input id ...;
        if id=3 or id=10 or id=23;
 
But we want a subset of a large number of ID numbers, which we've
got in a file, action.  The following works for a small test history
file, but fails for the large file because creating the data set history
fills /tmp.  We're running SAS 6.12 on Solaris.
 
        data action;
        infile action;
        input id;
 
        data history;
        infile history;
        input id ...;
 
        data subset;
        merge action (in=in_a) history (in=in_h);
        by id;
        if in_a & in_h;
 
Yes, we could make /tmp larger or use the WORK option to
specify a filesystem with more free space, but surely there
is some way to create the subset as the large file is read.
No?
 
Thanks,
 
Ron Stanonik
stanonik@nprdc.navy.mil
--
Ron Stanonik
stanonik@nprdc.navy.mil