|Examples:||Sorting by the Values of Multiple Variables|
You can specify only one collating-sequence-option and multiple other options in a PROC SORT step. The order of the two types of options does not matter and both types are not necessary in the same PROC SORT step.
proc sort data=mydata SORTSEQ=ASCII;
Linguistic collation is not supported on platforms VMS on Itanium (VMI) or 64-bit Windows on Itanium (W64).
When BY processing is performed on data sets that are sorted with linguistic collation, the NOBYSORTED system option might need to be specified in order for the data set to be treated properly. BY processing is performed differently than collating sequence processing.
The section on Specifying Linguistic Collation in SAS National Language Support (NLS): Reference Guide for detailed information.
The http://www.unicode.org Web site for the Unicode Collation Algorithm (UCA) specification.
Options can include one collating-sequence-option and multiple other options. The order of the two types of options does not matter and both types are not necessary in the same PROC SORT step.
The DUPOUT= and UNIQUEOUT= options are not compatible and cannot be specified simultaneously.
If the DUPOUT= data set name that is specified is the same as the INPUT data set name, SAS will not sort or overwrite the INPUT data set. Instead, SAS will generate an error message. The FORCE option must be specified in order to overwrite the INPUT data set with the DUPOUT= data set of the same name.
The EQUALS | NOEQUALS procedure option overrides the default sort stability behavior that is established with the SORTEQUALS | NOSORTEQUALS system option.
The EQUALS option is supported by the multi-threaded sort. However, I/O performance might be reduced when using the EQUALS option with the multi-threaded sort because partitioned data sets are processed as if they consist of a single partition.
The NOEQUALS option is supported by the multi-threaded sort. The order of observations within BY groups that are returned by the multi-threaded sort might not be consistent between runs.
In-database sorting occurs when the NODUPKEY option is specified and the system option SQLGENERATION= is assigned a DBMS and the system option SORTPGM=BEST.
Options NODUPKEY and NOUNIQUEKEY are not compatible. If these options are specified together, an error will be printed to the SAS log.
The DUPOUT= option can be used with the NODUPKEY option. It cannot be combined with the NOUNIQUEKEY option.
You can use data set options with OUT=.
The OVERWRITE option has no effect if you also specify the TAGSORT option. You cannot overwrite the input data set because TAGSORT must reread the input data set while populating the output data set.
The OVERWRITE option is supported by the SAS sort and SAS multi-threaded sort only. The option has no effect if you are using a host sort.
Using the PRESORTED option with ACCESS engines and DBMS data is not recommended. These external databases are not guaranteed to return observations in sorted order unless an ORDER BY clause is specified in a query. Generally, physical ordering is not a concept that external databases use. Therefore, these databases are not guaranteed to return observations in the same order when executing a query multiple times. Physical order can be important for producing consistent, repeatable results when processing data. Without a repeatable data retrieval order, PROC SORT does not guarantee the return of observations in the same order from one PROC SORT execution to another, even when the EQUALS | NOEQUALS option is used to request sort stability. Without a repeatable retrieval order, the detection and elimination of adjacent duplicate records by PROC SORT can also vary from one PROC SORT execution to another.
For information about the SAS system options REALMEMSIZE and MEMSIZE, see the SAS documentation for your operating environment.
If a failure occurs when adding the THREADS | NOTHREADS procedure option using the SPD engine, PROC SORT stops processing and writes a message to the SAS log.
The THREADS system option is honored if PROC SORT determines that multi-threaded processing is deemed to be beneficial. If the value of the SAS system option CPUCOUNT=1, then multi-threaded processing is not beneficial. However, you can specify the PROC SORT THREADS option to force multi-threaded processing when the system option is set to NOTHREADS or when the system option is THREADS and the procedure option is NOTHREADS. This option combination prevents multi-threaded processing and overrides the actions taken that are based on the system options. Note that when multi-threaded sorting is in effect and NOEQUALS is specified, observations within BY groups might be returned in an unpredictable order.
If multi-threaded SAS sort is being used, the UTILLOC= system option will affect the placement of utility files. Thread-enabled SAS applications are able to create temporary files that can be accessed in parallel by separate threads. For more information, see UTILLOC= .
The TAGSORT option is not supported by the multi-threaded sort. Specifying the TAGSORT option will prevent multi-threaded processing.