DATAn naming scheme and rollover

[gabonzo]

I use the _DATA_ trick a lot, especially for temporary datasets that I create inside macros and then delete when the macro has been executed.

I have found a problem with that though: it seems to me that the number 'n' in the DATAn naming scheme rolls over at some point, and n starts over from 1.

This causes a lot of problems, and my programs stop working.

I don't see in the documentation when the rollover is, does anybody know:

1. What the maximum allowed number for DATAn is
2. If it is possible to extend it

Thanks

[ChrisHemedinger]

I don't have an answer for your exact question, but I do have some advice from my experience. When I'm creating a series of temp/scratch data sets that I'm going to turn around and delete at the end of a process, I like to isolate them to their own folder (outside of WORK library). That gives me more options for deleting the whole shebang without worrying about conflicts.

I documented this in How to delete SAS data sets.  Specifically the trick I use is this:

/* create a subfolder in WORK for my data */
options dlcreatedir;
libname scratch "%sysfunc(getoption(WORK))/scratch";

/* just adding some data for example purpose */
data scratch.a1 scratch.a2 scratch.a3;
 set sashelp.class;
run;

/* more operations with scratch data */

/* now clear it when ready */
proc datasets lib=scratch kill nolist;
quit;

[gabonzo]

I like your idea, but it would be tricky to use it to solve my problem.

I don't create that many datasets per macro (perhaps three at most), but I have many macros, and in the current task I am doing a statistic simulation where I have to run the same set of macros over and over again for tens of thousands of times.

One way I could use your trick is to create a subfolder for each macro, but I am concerned the overhead will slow down an already slow program. Moreover, isn't PROC DATASETS supposed to be a bit of a resource hog?

Thanks anyway

[ChrisHemedinger]

If you don't like PROC DATASETS (it does have some overhead), then PROC DELETE is quick and light. You can delete a range, but not a "wildcard" list.

proc delete library=work data=DATA1-DATA5; run;

[Reeza]

Not a solution but similar to @ChrisHemedinger I use a common prefix, typically _dataSetName for all my temp tables and then drop them at once at the end using PROC DATASETS and the shortcut colon.

proc datasets lib=work;
delete _: ;
run;quit;

Here is a reference that illustrates how to refer to variables and datasets in a short cut list:
https://blogs.sas.com/content/iml/2018/05/29/6-easy-ways-to-specify-a-list-of-variables-in-sas.html

---

@gabonzo wrote:

I use the _DATA_ trick a lot, especially for temporary datasets that I create inside macros and then delete when the macro has been executed.

I have found a problem with that though: it seems to me that the number 'n' in the DATAn naming scheme rolls over at some point, and n starts over from 1.

This causes a lot of problems, and my programs stop working.

I don't see in the documentation when the rollover is, does anybody know:

1. What the maximum allowed number for DATAn is
2. If it is possible to extend it

Thanks

---

[gabonzo]

Ok, it seems a consensus is forming. Perhaps I will go with this solution.