Scot, Thanks for the info. I will try your code out to verify the result, but before I do that, will your code (SAS and R) work with variable names that are longer than 8 characters long without truncating the variable name in R? Also, I wonder about using your method or the PROC EXPORT method with larger data sets. The data sets I will be working with for the most part will not be that large, but for larger data sets, I do not like to create the intermediate csv file that PROC EXPORT creates because I do not like having unused files stored on my computer. Thanks, Stephen
--- Scot W McNary <smcnary at fellspt.charm.net> wrote:
Stephen,
I use SAS a lot and have to admit I use the proc
export version to csv
format you posted on a regular basis, however, I was
able to get the
transport version to work by using the xport engine
in a libname statement
as follows (SAS 8.2 on Windows98SE, rw1061):
# using SAS
libname check xport 'e:\testing.xpt' ;
data a;
do i = 1 to 10 ;
x = 1 + i ;
y = 50/x ;
output;
end;
run;
data check.a ;
set a ;
run;
proc print data = check.a ;
run;
The SAS System 15:31 Tuesday, December 24,
2002 1
Obs I X Y
1 1 2 25.0000
2 2 3 16.6667
3 3 4 12.5000
4 4 5 10.0000
5 5 6 8.3333
6 6 7 7.1429
7 7 8 6.2500
8 8 9 5.5556
9 9 10 5.0000
10 10 11 4.5455
# now reading into R
library(foreign)
test<-read.xport("e:/testing.xpt")
test
I X Y 1 1 2 25.000000 2 2 3 16.666667 3 3 4 12.500000 4 4 5 10.000000 5 5 6 8.333333 6 6 7 7.142857 7 7 8 6.250000 8 8 9 5.555556 9 9 10 5.000000 10 10 11 4.545455 The tricky part for me is having to suspend the metaphor of a "libname as folder" and think of the libname check as pointing to a file: "testing.xpt". It works that way in trying to read spss files with the spss engine too. I would assume it's also true for using bmdp and osiris engines, but I haven't had occasion to use them. Hope this helps, Scot -- Scot W. McNary email:smcnary at charm.net