SAS data sets are stored in a special format. Only software that can recognize the special format can read SAS data files. There are two principal types of SAS data sets for PC's. The first type is created by SAS for DOS using SAS version 6.03 or 6.04 and is characterized by files that have the suffix *.ssd. The second type is created by SAS for Windows using Windows-SAS Version 6.08 (or later) and is characterized by files that have the suffix *.sd2. The manner in which data are stored differs between the two types. As a result, the type of SAS data set has to be specified when reading a SAS data set.

The manner in which SAS data sets are stored in Windows-SAS differs from that in SAS for DOS. The data sets saved using the SAS version 6.08 format or higher have additional features that were not included in the DOS-SAS data sets. The additional features are designed to speed processing, and include

• "index" variables
• the ability to create "views"

We will not discuss these features here. DOS-SAS can not create data sets that have the Windows-SAS features. However, Windows-SAS can create DOS-SAS data sets. When reading a SAS data set, the SAS system needs to identify the type of SAS data set. It does so by identifying with the data set a SAS version, called an "engine".

There is an "engine" for each version of SAS. However, since the DOS versions 6.03 and 6.04 of SAS create identical data sets, and the Windows versions (6.08 or later) create nearly identical data sets, we can distinguish between two principal engines. The Engines are:

• V603/V604
  • Equiv. to DOS-SAS 6.03 or SAS 6.04, with data stored as *.ssd
• V608/V610/V612
  • Equivalent to Windows-SAS 6.08/6.10 and stored as *.sd2

While the version of SAS used to create a SAS data set is needed to read in a SAS data file, SAS data files using earlier SAS versions can be created using newer SAS engines.

SAS data files with the suffix *.ssd or *.sd2 do not uniquely identify the engine that created the data file, but are sufficient for SAS to read the data set. Engines are identified by specifying the "engine" option in a LIBNAME statements. For example,

LIBNAME old4 V604 ‘c:\temp\old';

will associate with the LIBREF "old4" the engine "V604". Thus, files saved in C:\TEMP\OLD with SAS versions 6.03 or 6.04 can be accessed using the LIBREF "old4".

More than one LIBREF can be assigned to the same path if SAS data sets that have been created with different engines are stored in the same location. For example, we may have an additional LIBREF defined for SAS 6.10 data sets given by :

 LIBNAME old10 V610 ‘c:\temp\old';

The LIBREF "old10" would be used to identify SAS data sets that were stored using SAS 6.10.

Note that the "engine" is an option in a LIBNAME statement, and not essential. When no "engine" is specified, SAS 6.10 will determine the "engine" it considers best to use. The engine is identified as follows:

•  a. if there are no SAS data files in the LIBREF, then the engine is the current SAS version executing, V612 (or later).
• b. if SAS data files exist in the LIBREF that were created by the same engine, then that engine is assigned the LIBREF.
• c. if SAS data files exist in the LIBREF that were created using different engines, then the most recent SAS engine will be used (ie. V612). 

Once an "engine" is associated with a LIBREF in a LIBNAME statement, all files read from or written to the LIBNAME location are read or written using the "engine" assigned to the LIBNAME.

( see Lecture 12)