5# pobel
但是point i即便已经停留在最后一条记录了，当_n_变为2时，它还会再次重新回到b的第一条记录的..无限循环哦..

3# soporaeternus
为什么叫大大....他真的好厉害...佩服....程序写的很漂亮...貌似什么模块都懂的...

召唤hopewell大大~

你们给他发个短信不就行了？

1) A usual SAS data step does more than it looks. For example, it implements a internal loop of observations so that one only needs define an operation for an observation. It has an output statement 'slightly' above a run statement. If one uses put _n_ one will see the internal loop counts.
But if one put a set statement explicitly into a do loop, then the internal loop will be overwritten. This sort style has some advantages if one fully understans the guts of SAS data step.

*****explicit loop****;
data repeat;
do until  (last);
set b end=last;
output;
put _n_=;
end;
run;

****a usual way in SAS books;

data repeat;
set b;
put _n_=;
run;

2) The point option on a set statement instructs a direct access to a data set. It provides some advantages. For example, one can create a reverse order data set by reversing points (do i=nobs to 1 by -1).  Or one may want to read obs in { 1, 101, 201, ...}. A direct access is usually slower than a sequential access.  Under a direct access one needs to tell SAS/data step when to stop.

data repeat;
do i=1 to nobs;
set b  point=i nobs=nobs;
output;
end;
stop;
run;