topic Re: sequentially designed datastep run codes within proc cas do not wait for its predecessor to fini in SAS Programming

sequentially designed datastep run codes within proc cas do not wait for its predecessor to finish.

acordes — Thu, 20 Mar 2025 18:21:01 GMT

I get bad results because they don't wait for each other.

The result is different from the one I egt when I run both datasteps one after another.

proc cas;
function doesTableExist(casLib,casTable);
	table.tableExists result=r status=rc / caslib=casLib table=casTable;
	tableExists = dictionary(r, "exists");
	return tableExists;
end func;

source myDataStep;
	data casuser.origin_filtered_by_time;
		set &fname.;
		if timestamp >= input("&last_start_date.", datetime20.);
		keep timestamp var1; 
	run;
endsource;

source myDataStep1;
	data casuser.result_groups_temp / sessref="mySession"
		single=yes;
		set casuser.origin_filtered_by_time;
		by timestamp var1; 
		retain group_id;
		if lag(var1) ^= var1 then group_id + 1;
		if first.timestamp;
	run;
endsource;



/* check whether the process has been carried out at least once */
tableExists_watchdog = doesTableExist("&liby", "&casy_watchdog");

/* in that case, proceed with the following steps */
if tableExists_watchdog ^=0 then do;

	dataStep.runCode result=r status=rc /
		code=myDataStep;
	dataStep.runCode result=r status=rc /
		code=myDataStep1;


run;

Re: sequentially designed datastep run codes within proc cas do not wait for its predecessor to fini

SASJedi — Mon, 24 Mar 2025 14:15:15 GMT

In what way is the result different from your expected result?

Re: sequentially designed datastep run codes within proc cas do not wait for its predecessor to fini

acordes — Mon, 24 Mar 2025 17:19:06 GMT

The final output table had less entries.

It seemed to me that the second data step didn't wait for the first one to complete. As if it started setting the input table=output table from the first data step before it had all rows written to table.

Re: sequentially designed datastep run codes within proc cas do not wait for its predecessor to fini

SASJedi — Mon, 24 Mar 2025 19:55:05 GMT

In your second DATA step, this statement:

		if first.timestamp;

ensures that only the first record for each timestamp group is output. So the output data set from the second data step (casuser.result_groups_temp) will most likely have significantly fewer observations than the input data set (casuser.origin_filtered_by_time).

I emulated your process by loading the sashelp.cars dataset to my casuser caslib, and using it in a simplified form of your program. It all worked exactly as I would have expected...