It is possible to use DEVICE=ACTIVEX or JAVA as a goption so that the data points used to generate the graph and graph attribute (java object or active x control specific values). The drawback is that this provides access to the data points or even raw data through the object or in the raw html file to users. It also does not support all styles and options but there's typically a way around getting the desired output. The user does not need to have SAS installed as there is a prompt to download the appropriate activeX control from SAS website (default path) to view the graphs. Alternatively, if you work on a closed network, you could download a save of the activex EXE from sas.com and store it on a shared location and use an option (I forget which) to change the default path to prompt users lacking the activex control on their machine for download and install. Anyway, this is really the only alternative if you want to keep on using html. This is in the nature of HTML as Ballard pointed out to need to be able to access the image locations. ... View more

Throw options symbolgen; for debugging. According to your log, COLUMNLIST macro variable was never affected a value as the condition segment is in comments so you tried to define a variable-list array without providing any variables. This should be where your error comes from. I would also add a columnlist param to the the macro as DN pointed out. you can embed your condition within the macro or do a simple error handling. ... View more

Thanks everyone its working ... View more

So is the idea to perform an horizontal split (left most variables --> right most variables) on a dataset with &N representing the maximum number of variables per output dataset?  So if I had dataset HAVE with variables VAR1 to VAR500 I could split this into three files with max of 200 variables each by writing code like this: data split1(keep=var1-var200)        split2(keep=var201-var400)        split3(keep=var401-var500) ; set have ; run; If the names are instead arbitrary instead of a simple numbered range of names you can use the -- range specification instead. So if you wanted to split SASHELP.CLASS with a max of three variables per the resulting split would be NAME--AGE and HEIGHT--WEIGHT. So what is a reasonable way to generate that code. I find that this is easier with a data step as you can take advantage of the FIRST./LAST. processing to generate the code. %macro hsplit(inds,nvars,prefix=split); proc contents data=&inds out=varlist(keep=name varnum) noprint ; run; proc sql noprint ;   create table varlist as   select int((varnum-1)/&nvars)+1 as group        , varnum        , name   from varlist   order by 1,2   ; quit; filename code temp; data _null_;   set varlist end=eof;   by group ;   file code ;   if _n_=1 then put 'data ';   if first.group then put "  &prefix" group ' (keep= ' name '-- ' @ ;   if last.group then put name ')' ;   if eof then put ';' / "  set &inds;" / 'run;' ; run; %inc code/source2 ; %mend hsplit; %hsplit(sashelp.class,3) %hsplit(sashelp.cars,7) ... View more

You can generate the code from the list of variables. data have; input emp_code  $ li_Jan Li_feb MF_Jan MF_feb GI_jan GI_feb ; cards; a 2 5 5 10 9 8 b 3 6 6 6 6 2 c 1 3 2 3 7 2 d 4 4 1 1 2 1 run; proc sql noprint ;   create table varlist as     select upcase(scan(name,-1,'_')) as month          , name     from dictionary.columns     where libname='WORK' and memname='HAVE'       and calculated month in           ('JAN','FEB','MAR','APR','MAY','JUN','JUL','AUG','SEP','OCT','NOV','DEC')     order by 1,2   ; run; filename code temp; data _null_ ;   set varlist ;   by month ;   file code ;   if first.month then put month +(-1) '_Sales=sum(0' @;   put ',' name @ ;   if last.month then put ');' ; run; data want ;   set have ; %inc code / source2 ; run; 288  data want ; 289    set have ; 290  %inc code / source2 ; NOTE: %INCLUDE (level 1) file CODE is file .../#LN00022. 291 +FEB_Sales=sum(0,GI_feb ,Li_feb ,MF_feb ); 292 +JAN_Sales=sum(0,GI_jan ,MF_Jan ,li_Jan ); NOTE: %INCLUDE (level 1) ending. 293  run; NOTE: There were 4 observations read from the data set WORK.HAVE. NOTE: The data set WORK.WANT has 4 observations and 9 variables. ... View more