I use an LRECL of 2048 and haven't had problems. The important thing is to add record_separator=none to the ODS tagsets.excelxp statement. Otherwise, you will get mysterious failures. ... View more

I agree about the silliness of not allowing the STATUS command - but lots of places practice security theatre - just for show, no real value - so I'm not surprised it happens. ... View more

Found this somewhere on the intertubes: ===== DATA ZZGETUSR (Drop=TIOTPTR JSCBPTR SSIBPTR JCTPTR ACTPTR Label='User information on MVS job'); LENGTH PROGRAMR $20 JOBNAME $8 PROCSTEP $8 STEPNAME $8 PROGNAME $8 JOBNUMBR $8 USERID $8 ; TIOTPTR = peek(peek(540) + 12); JOBNAME = peekc(TIOTPTR, 8); PROCSTEP = peekc(TIOTPTR+8, 8); STEPNAME = peekc(TIOTPTR+16, 8); JSCBPTR = peek(peek(540) + 180); PROGNAME = peekc(JSCBPTR + 360, 8); SSIBPTR = peek(JSCBPTR + 316); JOBNUMBR = peekc(SSIBPTR + 12, 8); JCTPTR = peek(JSCBPTR + 260); JOBCLASS = peekc(JCTPTR + 47, 1); MSGCLASS = peekc(JCTPTR + 22, 1); ACTPTR = INPUT('00'X || peekc(JCTPTR + 56, 3), PIB4.); PROGRAMR = peekc(ACTPTR + 24, 20); SYSTEMID = PEEKC(PEEK(peek(16, 4) + 196, 4) + 16, 4); USERID = peekc(peek(peek(548) + 108) + 192, 8); put (_character_) (=/); run; ===== ... View more

I was thinking that if SAS is not suitable for his/her needs, s/he should find another software solution. Or perhaps some classes would be helpful. ... View more

I wouldn't say that SAS is always relational, but it can be relational, just as Oracle can be relational (or not). Saying that a relational database can't be implemented in SAS (especially if you have some of the newer SAS products like TSPL) would be overreaching. Saying that all Oracle or DB2 or (fill in commonly used commercial database name here) databases are relational would be just as much of a stretch. In other words, it's the design that makes something relational, not (just) the platform. ... View more

What you mean when you say that the SAS database architecture is not truly relational? You can create fifth normal form databases with indexes and constraints in SAS, and you can create completely denormalized databases in Oracle. So what is the essential difference? ... View more

My guess (an informed guess, but still a guess) is that Excel switches to "this is a text file" behavior when it encounters a line that's not defined by the SYLK specification. I occasionally see the same behavior with XML files if somethings not quite right. The Content-Type and Content-Disposition are there because I was creating these files in SAS/Intrnet. ... View more

If you are storing it in SAS as a number, you can display it with leading zeros using the z5. format (In Excel, you can use an Excel format to display ZIP codes with leading zeros). If you are storing the number in a 5-byte character field in SAS, you could put the zeros back with the translate function (first make sure the remaining digits are right-justified). ... View more

When you double-click on an XML file created by ExcelXP, Excel isn't opened directly. Instead, the XML file is opned by another program, MSOXMLED, which scans the file looking for more specific information on what to do next, such as "". If it finds that line, it passes the file name to Excel; if it finds another program name, it uses that; and if it doesn't find anything that can help it decide what program to use, it opens the XML file for viewing in Internet Explorer. A SYLK file, on the other hand, is opened directly by Excel. It's a very old file format. If I run the code in your example and navigate to c:\temp in the SAS Explorer, I see an Excel icon for the file, and double-clicking on it opens Excel. I'm not sure what you're doing when you say that the SAS Results Viewer shows you an XML icon. Perhaps if you went into the SAS registry editor and set up an entry for the SYLK filetype, SAS would handle it correctly. I don't know. That tagset was written to address a specific need, and as a starting point to learn about tagsets. I was unable to find documentation explaining all the options and inter-relations between the different parts of a SYLK file, so the results are very rudimentary. Now that XML is available, SYLK is probably not the best choice for getting data into Excel (unless, of course, you're still using Excel 5 or below). ... View more

It will be easier to unconditionally create the error data set, and halt execution if the data set contains observations. ===== data _null_; set errors; abort abend; run; ===== Will abort if there are errors, and continue merrily along with the error data set has observations. ... View more

While it is true that PROC TABULATE seems designed for summary reports, it will effectively create a detail report if every observation has a unique combination of CLASS variables. You can create a pseudo-OBS variable to ensure uniqueness (preferably in a view to save on I/O). ... View more

You didn't say how you are creating the output. If you're using PROC TABULATE, use the NOCONTINUED option in the TABLE statement. ... View more

I don't see a need for the two sorts in LAP's code. ... View more

Here's one way: ===== data test; input a b; cards4; 1 23 2 15 3 45 4 55 5 67 ;;;; options mergenoby=nowarn; data new; retain a b c; merge test test (keep=b rename=(b=c) firstobs=2) end=eod; if eod then c = 0; run; options mergenoby=error; proc print; run; ===== Sorry about the lack of indentation - the forum software doesn't preserve it. ... View more

SQLOBS is set by PROC SQL. It's set even if you're not reading from an external database at all. SQLOOPS is also set, but I have idea how PROC SQL sets the value. ... View more