The behavior you're seeing may have more to do with IE than it does with SAS.  In my IE, different HTML results from SAS go into tabs not separate IE windows.    Which SAS interface are you using?  How are you opening IE?     Jim ... View more

Hi, @AnnaBrown,   As I recall, I did use full screen, and I have a monitor attached to my laptop, so it's a fairly large display, but if no one else experienced it, then it must be something unique to my set up.   I look forward to September's presentation,   Jim ... View more

Probably (probably, not certainly), you're not getting an execution log because the job never got to the point of execution.  The job failed before execution.   If a job works for one person but not for another, it's often either: A differences in permissions (access levels such as read vs. write or being allowed to read at all, for example). A difference in environments (does the second person have the same drives mapped, for example). My response is a little bit general, but that's where I would start looking (at the permissions and at any potential differences in the environment).   Jim ... View more

What is the source of the elapsed times?    Are you getting it from outside SAS?  If it's from outside SAS (Excel for example), the representation can be different and a conversion is necessary.  Again using Excel as an example, Excel bases everything off of Jan 1, 1900.  SAS, on the other hand, bases everything off of Jan 1, 1960.  Excel date-times will appear to be far in the future in SAS if not converted.   Inside SAS, normally one just subtracts the ending date-time value from the starting date-time value.  I typically use the format TIME13.2 with elapsed times.   As an example: %LET Start_Datetime = %SYSFUNC(INPUTN('07sep2018:23:00:00'dt, DATETIME18.)); %LET End_DateTime = %SYSFUNC(INPUTN('08sep2018:01:00:00'dt, DATETIME18.)); %PUT %SYSFUNC(COMPRESS(%SYSFUNC(PUTN(%SYSEVALF(&End_DateTime - &Start_DateTime), TIME13.2)))); Results in: 2:00:00.00 Indicating that 2 hours, 0 minutes, and 0 seconds have elapsed.   Jim ... View more

Can you tell us the source of the column?  Will it be from another SAS dataset?  Is it based on a calculation (as in @PaigeMiller's response)?  Is it from a file?  A database?  How will you obtain this value?    Adding a new column to a SAS dataset is pretty easy.  Say you have an existing data set, let's call it "Have."  If you want to add a column to the data, one way is a Data step, for example, the below.  In my existing data, if I have a column named "Centimeters" but I want a new column called "Inches," I can create a new column by using the standard conversion formula as shown. Data Want; Set Have; Inches = Centimeters / 2.54; Run; The output of the above Data set will be a new dataset named "Want".  The new data set will be a copy of Have but will have a new column added to, a column named "Inches."   Jim   ... View more

@raj712 wrote: I haven't passed anything as -MSG. You personally may not have, but it's in your configuration. UNABLE TO OPEN/READ MESSAGE FILE USING THE -MSG PATHS Jim ... View more

@Patrick,    Good approach.    If I may ask:  Why use multidata:'y'?  It sounded as though we wanted a unique count.  Wouldn't multidata:'y' permit duplicates?   Jim ... View more

@Alexxxxxxx,    I would agree with @Patrick that if there is a way you can get all the data you need without generating so many rows, that will be your best method of reducing run time.   One thing I notice, is that it looks like the Data step really isn't necessary.  A Case statement in the first SQL procedure appears to produce the same results.  Check it and see what you think.  This would eliminate having to read every row again in the Data step.   Here's the modified SQL (added Case statement): PROC SQL; CREATE TABLE want1_&year. AS SELECT DISTINCT citing1.Applicant_name, citing1.appln_filing_year, citing1.appln_id, pre_patent, citing1.cited_appln_id, citing2.appln_id_citing_p5 ,CASE when cited_appln_id=appln_id_citing_p5 then 1 ELSE . END AS cited_patent FROM have1_&year. (keep= Applicant_name appln_filing_year appln_id cited_appln_id pre_patent)as citing1 JOIN have2_&year.(keep=appln_id_citing appln_id_citing_p5) AS citing2 ON citing1.appln_id = citing2.appln_id_citing order by citing1.appln_id ; QUIT;   Results:   ... View more

Can you post your log?  Exactly which step is it that is taking a long time?  The 2nd SQL step?   How many rows does Want1 wind up having?   The first thing that occurs to me is that a WHERE clause will be more efficient than a HAVING with a summary function.   Something like: PROC SQL; CREATE TABLE want3_&year. AS SELECT DISTINCT Applicant_name, appln_filing_year, appln_id, cited_appln_id, pre_patent, cited_patent FROM want2_&year. WHERE cited_patent > 0 ORDER BY /* GROUP BY*/ Applicant_name,appln_id,cited_appln_id /* Having cited_patent=max(cited_patent) */ ;   However, if there are cases where the max(cited_patent) is going to be zero or missing and you're going to want to have them in the final output, my WHERE suggestion won't work.   Jim ... View more

Hmmm.  Good question.  Not my area of expertise.   I wonder if either @gwootton or perhaps @SASKiwi might be able to shed some light on this.   Jim ... View more