@sas_nutzer wrote: Thank you for the answer, but the source says: "The RETAIN statement is most often used to reorder variables because no other variable attribute specifications are required. The RETAIN statement has no effect on retaining values of existing variables being read from the data set."   So it shouldn't behave like that, should it? The important part is this: of existing variables being read from the data set Since dummy is not being read from a dataset, it is affected by the RETAIN statement. Although existing variables (those read by a SET or MERGE) are retained by definition, they are always overwritten when a new observation is read. Dummy is not overwritten and therefore keeps its value. ... View more

You're right, it's not the log, it's the results.   @Ronein use ODS statements to turn off all output first, then open a single destination to a file. ... View more

Use PROC PRINTTO to redirect the log to a file. Then you can inspect the file to find out what causes this large log. ... View more

Please ask your question in a new thread. ... View more

3) Macros are stored globally. Only macro variables have scope. ... View more

@Feefee wrote: Hi,    I have dataset that looks like this:    No, you don't. SAS names can't have blanks or linefeeds in them.  Please do ALWAYS post example data as working DATA step code with DATALINES. ... View more

My first idea would be that Snowflake counts UTF characters, while SAS counts bytes. ... View more

@quickbluefish wrote: Yes - in this case, you can also just do: data want; set have; by custid; seq+first.custid; run; Kudos for brevity. I love tight code. ... View more

All the more reason to conduct a real-life test. ... View more

Rather than tinkering around with PROC SQL, you should try DATA step methods, e.g. hash objects if data can be fit into memory. ... View more

Maxim 4. Try it and compare the results. ... View more

This is for the data you posted: data want; length tbl1 tbl2 $8; call missing(of tbl:); merge have (where=(tbl="tbl1") rename=(field=tbl1)) have (where=(tbl="tbl2") rename=(field=tbl2)) ; drop tbl; run; For more values of tbl, run a preliminary step to get the list of values, and use macro %DO loops to create the code dynamically: %macro want; %local tables tbl i; proc sql noprint; select distinct tbl into :tables separated by " " from have; quit; data want; length &tables $8 ; call missing(of tbl:); merge %do i = 1 %to %sysfunc(countw(&tables)); %let tbl = %scan(&tables,&i); have (where=(tbl="&tbl.") rename=(field=&tbl.)) %end; ; drop tbl; run; %mend; %want ... View more

Doesn‘t matter. DICTIONARY.COLUMNS has the metadata of all currently assigned libraries. ... View more

Why do you have numeric zipcodes in the first place? This has to be a faulty import process. ... View more

It‘s already there: SASHELP.VCOLUMN (or DICTIONARY.COLUMNS in PROC SQL). ... View more