Why are two datasets created here instead of 1 ?

[Gemie]

%macro nmtest(a1=);

data class_&a1.;
   set sashelp.class;
run;

%mend nmtest;
%nmtest(a1=%str(bb));

[PaigeMiller]

Your use of %str() is not needed here.

%macro nmtest(a1=);
data class_&a1.;
   set sashelp.class;
run;

%mend nmtest;
%nmtest(a1=bb);

This produces only one data set.

If you turn on macro debugging ... via this command

options mlogic;

and then re-run your code, you will see in the log that &a1 has additional characters introduced by the unnecessary %str( )

118   options mlogic;
119   %macro nmtest(a1=);
120   data class_&a1.;
121      set sashelp.class;
122   run;
123
124   %mend nmtest;
125   %nmtest(a1=%str(bb))
MLOGIC(NMTEST):  Beginning execution.
MLOGIC(NMTEST):  Parameter A1 has value 
bb
SYMBOLGEN:  Macro variable A1 resolves to bb
SYMBOLGEN:  Some characters in the above value which were subject to macro quoting have been unquoted for printing.
MPRINT(NMTEST):   data class_bb;
MPRINT(NMTEST):   set sashelp.class;
MPRINT(NMTEST):   run;

Perhaps @Astounding might be able to explain in more detail

[Gemie]

Thanks. I was only expecting class_bb to be generated and not class_ .

[Kurt_Bremser]

The %str function inserts a special macro token wich causes the dataset name to appear as two names for the data step compiler.

[Quentin]

Your expectation is correct. It's a small bug that has been around for a long time that sometimes invisible quoting characters can cause problems for the compiler when SAS fails to automagically remove them. As Tom showed, in those cases explicitly %unquoting() the value is usually a solution.

[PaigeMiller]

---

@Quentin wrote:
Your expectation is correct. It's a small bug that has been around for a long time that sometimes invisible quoting characters can cause problems for the compiler when SAS fails to automagically remove them. As Tom showed, in those cases explicitly %unquoting() the value is usually a solution.

---

But not using %str() in the first place also eliminates the problem. Since &a1 is going to be part of a data set name, there's no need for %str() at all, there cannot be special characters in the data set name that %str() has to mask.

[Tom]

The macro quoting is confusing the parser.  The macro quoting on the macro variable causes the macro processor to see that as two tokens which causes the SAS processing to also see it as two.

You can remove the quoting using %UNQUOTE() macro function.

248  %macro nmtest(a1=);
249
250  data %unquote(class_&a1.);
251     set sashelp.class;
252  run;
253
254  %mend nmtest;
255  options mprint;
256  %nmtest(a1=%str(bb));
MPRINT(NMTEST):   data class_bb;
MPRINT(NMTEST):   set sashelp.class;
MPRINT(NMTEST):   run;

NOTE: There were 19 observations read from the data set SASHELP.CLASS.
NOTE: The data set WORK.CLASS_BB has 19 observations and 5 variables.
NOTE: DATA statement used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds