On Tue, 15 Dec 2009 13:41:28 +0100, js8765 <js8765@GOOGLEMAIL.COM> wrote:
....
>I'm learning SAS at the moment and in general I find the documentation
>provided by SAS to be very thorough and easy to follow. However, I feel
>I would be able to learn a lot quicker and understand the different
>aspects better if I could find more information on what SAS is actually
>doing behind the scenes in each case.
>
>For example, when I eventually found information on the way SAS
>processes match-merge operations (i.e. information on how the
>observation pointer moves through the different data sets, etc.),
>suddenly everything became clear and it was easy to predict what would
>happen in different situations.
>
>Another example is the UNIQUE option used when modifying data using an
>index. It finally made much more sense when I discovered somewhere that
>the data set normally only returns to the top of the index when the
>value of key variables changes between execution of the statement with
>the KEY= option.
....

Hi, js:
Both of your examples involve how the data step and more specifically how
the Set, Merge, Update, and Modify statements work.
In addition to reading the sas manuals, I recommend reading Howard
Schreier's brilliant SUGI 30 paper, "Let Your Data Power Your DATA Step:
Making Effective Use of the SET, MERGE, UPDATE, and MODIFY Statements,"
(available from his web site at http://www.howles.com/saspapers/).
Using the uniqueness of PDV variables and the novel concept of resetting
boundaries, he gives readers a general and coherent framework of
understanding the behavior of the data step and the SMUM statements.
Hope this helps a bit.
Cheers,
Chang