The annomac %RECT is calling for the POLYLINE, which I believe has been replaced with the POLY function. I do not know what the interaction is, but replacing the macro call with its resultant code and replacing POLYLINE with POLY seemed to clear up the problem on my test. the code snippet becomes: [pre] %*rect(x1,y1,x2,y2,H05A66AA,1,2); /*IF "H05A66AA" =: '*' THEN ;*/ /*ELSE */ color = "H05A66AA" ; LINE = 1; SIZE = 2; x=x1; y=y1; function="POLY "; output; x=x2; y=y1; function="POLYCONT"; output; x=x2; y=y2; function="POLYCONT"; output; x=x1; y=y2; function="POLYCONT"; output; x=x1; y=y1; function="POLYCONT"; output; [/pre] Another reason I tend to shy away from the %annomac macros. ... View more

You say that you have 150 'files'. This data step is processing one file at a time, and it creates one data set at a time. For a given file what do you want to have happen? This DATA step creates a single observation data set under certain conditions. Do you want all 150 files to be concatenated into a single data set - with this 'special condition' being just a row in that data set? ... View more

Ron Fehd has written several papers on list processing in the macro language. a couple of starting points might be: http://www.sascommunity.org/wiki/List_Processing_Basics_Creating_and_Using_Lists_of_Macro_Variables http://www.lexjansen.com/pnwsug/2009/Fehd,%20Ron%20and%20Art%20Carpenter%20-%20List%20Processing%20Basics.pdf ... View more

I am not sure what you are trying to do here. Is there only one incoming row? or do you want to issue the STOP only if there is no information? Let me take a guess and we can refine the answer: [pre] data CDM1.temp_&seed; infile "c:\CDM1\&commandfile..out" truncover; input test $21. @; if test="Information Criteria" then do; * this is good data - read it; input / / @40 num_fre_par / @40 akaike5 / @40 bayesian / @40 sam_siz_adj_bic ; output; end; * add the END; else do; * set values to missing and stop reading data; num_fre_par=0; akaike5=0; bayesian=0; output; stop; end; drop test; run; [/pre] ... View more

the open=defer causes the data sets to only be opened when they are ready to be read. Otherwise all 150 data sets would have to be open at one time (the default). Under some (most I would assume) OSs there are limits as to the number of open data sets. Using the colon operator in your original code allowed the data sets to be read as they were encountered, which was apparently not in named order. ... View more

As an aside. Peter shows what I believe is the correct way to calculate the Nth root, but the GEOMEAN also calculates this value. The following shows two ways to calculate the the 6th root of 64 (a number even I can double check). [pre] data a; x =64**(1/6); y = geomean(64,1,1,1,1,1); put x= y=; run; [/pre] ... View more

You need to review the difference between the routine CALL SYMPUTX and the use of %LET in a DATA step. It is not possible to assign DATA step variable's value to a macro variable using a %LET - the DATA step routine SYMUTX is used instead. It is very important to understand why. in your program (snippet shown here). [pre] data b; set data1; %let s=eq_price; * underlying security price; [/pre] the macro variable &S will contain the letters e-q-_-p-r-i-c-e, not the variable and not the value of the variable. This means that the %IF in the macro %P will always be false. ... View more

Take a look at the %ABORT statement. Depending on the level of termination that you want to achieve, it may be helpful. It also allows you to 'pass back' a return code which you may be able to use to determine your restart point. ... View more

here is a possible solution: [pre]%macro printit(dsn=); title1 Printing &dsn; proc print data=&dsn(obs=5); run; %mend printit; %macro getdatalist(lib=); %local count i; proc sql noprint; select distinct memname into :mem1 - :mem&sysmaxlong from dictionary.members where libname=upcase("&lib"); quit; %let count = &sqlobs; %do i = 1 %to &count; %printit(dsn=&lib..&&mem&i); %end; %mend getdatalist; %getdatalist(lib=sasclass) [/pre] ... View more

Just a couple of thoughts on your macro. 1) it is never wise to embed macro definitions. 2) there are a number of ways to count the number of obs without reading the whole data set. Although not the most classiest, consider: [pre] data _null_; call symputx('numobs', nobs); stop; set &dsetin nobs=nobs; run;[/pre] 3) the incoming data set &dsetin is being sorted multiple times including inside a macro call that is inside a macro loop. 4) did you consider the suggestions of Patrick and Ksharp? taking a big(?) data set breaking it up and putting it back together is always resource intensive. ... View more

Can you show us a couple of simple data sets that illustrate the problem and the hoped for result? It might help us understand what you are trying to achieve. ... View more

this post went unanswered from a coupleof weeks ago. Did you get the solution that you needed? If not are either of your data sets of a size that allow pre processing? ... View more

Richard_hu2003, SPR is absolutly correct but the reason that there is a local and global version of &A is more subtle. A slight change in the program will give D as the answer. [pre]%let a=cat; %macro animal(b=frog); %let a=bird; %mend; %animal(b=pig) %put a is &a; [/pre] now there is no version of &A on the local table, and cat is replaced by bird. In your original version of the macro %ANIMAL it is the parameter (A=) that forces a &A to appear on the local table. A %LET inside the macro is not sufficient. I feel that this is an important, albeit fairly subtle, difference in behavior that gets folks into trouble. ... View more

if you cannot use a list abbreviation such as suggested above, you can control the order using manually generated lists of variable names. There are macro language techniques for generating these list that have been discussed in other threads in this forum, but when it comes to typing I have little patience so an alternative to typing and an alternative to the macro approach is copy/paste, if you are using windows. To type without so much typing: 1) Use a proc contents to write the variables to the LISTING destination, consider using the POSITION option. 2) while holding the ALT key and the LMB drag the mouse down the list of variables. 3) copy and paste the list into a fresh editor window 4) rearrange as desired (I like to group variables by use - this makes life easier later when I want to build a VAR or CLASS statement. 5) paste list into LENGTH or RETAIN statements as appropriate (KEEP and DROP also but they do not determine variable order) Message was edited by: ArtC ... View more