Linlin, I am trying to reverse engineer your example. Ia ma seeing: - I small dataset with dates.   Could be optimized by using formats or hashing when needing a table lookup - A big dataset that is summarized by id's   Proc summary/means could be used to do that job.   With in-database processing there is no need for sorting I assume. When it is ordered but not sorted it will also work.     - The big dataset is merged with it is summarized by id's   And then I get lost as subsetting the dataset in several steps on count....    ... I got lost there on the intention... Please explain. - The resulted selections are concatenated again   Why losing the information of how it was constructed?     ... View more

Very nice, the idea of better performance limiting computer resource usage by using Formats is being got Having them defined could be a very handy one wiht using the full (SAS) dataset. The next level of this is Multi label formats. Using different classification / format recodings on the same data. Base SAS(R) 9.3 Procedures Guide, Second Edition ( Example 17: Using Multilabel Formats ) ... View more

as IIF is from thinking modifying data in the MS-access limitations... Now thinking in SAS oppurtunities... - If you have the format defined. You could possibly have no need to change data.   Many procedures (eg report/tabluate) are accepting the format being defined there and it works like the data was changed (but not really)   This approach can save a lot of time as it is  bypassing the need for data conversions ... View more

SAS(R) 9.2 National Language Support (NLS): Reference Guide ( specifying a Locale ) is telling: DATESTYLE=    DFLANG=    ENCODING=   PAPERSIZE=     TRANTAB= For the default values. SAS(R) 9.2 National Language Support (NLS): Reference Guide ( LOCALE= Values and Default Settings for ENCODING, PAPERSIZE, DFLANG, and DATESTYLE options) SAS(R) 9.2 Intelligence Platform: Web Application Administration Guide, Fourth Edition (SAS Web Infrastructure Platform) ... View more

SAS(R) 9.3 Macro Language: Reference (%sysfunc --  Example 5: Determining the Number of Variables and Observations in a Data) %let dsid = %sysfunc( open(myDataSet) ); %let nobs = %sysfunc( attrn(&dsid,nobs) ); %let rc = %sysfunc( close(&dsid) ); %if &nobs gt 0 %then %do. . . ... View more

The libraries are physical on the OS (Windows/Unix) directories. Creating a directory is mouse click- create-dir/map/folder (whatever naming is used)  or the batch commands like md (windows), mkdir(Unix). Aside SAS administration some OS en security administration is getting involved. For those parts possible (not necessary) other people are needed to get involved. Your SAS admin should be on that track. Havingg the physical location defined/ready you SAS admin should define the libnames and connect them to servers (SAS metadata). By that the librarie-names shoudl become available to all the users using a SAS server(service like SASapp). The datasets in the librarie will become visible and will be retrievable.  If this is your goal there is no need for further actions. Are you going to use create new data out of those datasets, using DI (data integration studio) , defining limiting views (information map studio), prompt using datasets as input you need to register the related tables into metadata. In this approach you are building up a more complex IT environment. ... View more