From SAS documentation:
As a general recommendation: use a DATA step rather than PROC APPEND to re-create a large data set from smaller subsets.

这里 DATA是例如：
set work.one work.two work.three;

PROC例如：
proc append base=work.orders data=work.one;
run;
proc append base=work.orders data=work.two;
run;
proc append base=work.orders data=work.three;
run;

我不明白的是：append 不是应该比set 更efficient么（因为前者不需要读work.orders)。所以不明白为什么书上这样说。

谢谢！