Trouble is MISSING/NULLS have different meanings for ORACLE and SAS. When you perform an Order by through SQL Pass-Through (on the ORACLE side), it seems NULLS are considered a very big number. When you perform a SAS SORT, MISSINGS are considered a very little number. Have you tried to order through a COALESCE function? ORDER BY COALESCE (COL, -1) The COALESCE function causes any results that are NULL to be replaced by some other value you wish (in the example, -1). Not a pretty solution, but a easy one that may do what you whish. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

> Thank You for your answer Daniel. > > But if you look at my post more closely, You would > see, that i use inherited library > and I use it successfully. > > Here is code: > rsubmit _01 wait=no macvar=JOB01 signonwait=y > persist=n; > data _null_; > set testine.Conf_action_types(obs = 10); > run; > endrsubmit; > > and here is the log output: > NOTE: Remote submit to _01 commencing. > > NOTE: There were 10 observations read from the > data set TESTINE.CONF_ACTION_TYPES. > NOTE: DATA statement used (Total process time): > real time 0.08 seconds > cpu time 0.02 seconds > > > NOTE: Remote submit to _01 complete. > NOTE: Remote signoff from _01 commencing. > NOTE: SAS Institute Inc., SAS Campus Drive, Cary, NC > USA 27513-2414 > NOTE: The SAS System used: > real time 10.53 seconds > cpu time 0.14 seconds > > NOTE: Remote signoff from _01 complete. Yes, but this doesn't mean it's officially supported. There are many SAS features available that are not yet fully supported. You probably won't get that error if you remove the testine. Scott is right, you should open a track to clarify this situation (if metalibs can be inherited or not, and if positive, why do you get this error). Check the FAQ's for Tracks here: http://support.sas.com/techsup/opentrackshelp.html Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

I suspect you're having trouble, because you're trying to inherit a "special" library (metadata engine) which is by definition a remote lib in the metadata server. So you're trying to remotely access a remote library. For example, inheritlib will not work if you try to remote signon form another remote session (but will work fine if from within a local session). Try to remove the testine from your script, and see if the problem persists. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

OK. I'll try not to give you that one free. 😄 They are several approaches for your question. I'll point only one. My understanding, is that you want to summarize your data by PERSON, but retaining the ADMISSIONs whenever there is an 'X' within the group. So you simply need to work within the group, retaining the value of each admission. the RETAIN statement will do precisely that. Hold the last variable value within each datastep iteration. See the online documentation: http://support.sas.com/documentation/cdl/en/lrdict/61724/HTML/default/a000214163.htm To summarize your data, you just need to work with the FIRST / LAST statement (which will indicate if the iteration is at the top, middle, or bottom of the group) and perform an explicit output when desired (here, tipically at the end). See the online docmentation here: http://support.sas.com/documentation/cdl/en/basess/58133/HTML/default/a001349233.htm Combine these two features in one datastep, and you'll have your problem solved. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Everything you pointed is possible through the Open Metadata Interface, of course with the right authorization level. Not an easy task though. You'll have to code your needs into the correct XML request. Check the online publishing, everything is in there (at least the important facts): SAS 9.1.3 Open Metadata Interface User's Guide: http://support.sas.com/documentation/onlinedoc/91pdf/sasdoc_913/omd_interug_9964.pdf Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Another approach would be to code a macro that will delete the dataset (name passed as argument) if it is indeed empty. You see, no datastep code will execute if the dataset is empty. So, if you try to assign a value to a macro var from within the datastep, this assignment will only occur if the dataset has at least one row. Assume that the dataset is empty, let say: %let EMPTY=1; Then try to reset the macro var within the datastep; data _null_; set &DATA; call symput('EMPTY','0'); /* this will only occur if &DATA is not empty */ run; Now, you just need to check the macro var for its value at the end, to perform (or not) a proc dataset delete, %if &EMPTY %then %do; proc dataset lib=&LIB nolist; delete &DATA; /* delete dataset */ quit; %end; Next, you just need to invoke the macro at every point where it is needed. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Could you please share more of your code (lib assignments, signon and rsubmit script), because I don't see anything wrong about it. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

You should override the default alignment for the put statement (in your case -R option). put - ; for example: put X $10 -R; /* x right aligned in a 10 char space */ Please, check online documentation here: http://support.sas.com/documentation/cdl/en/lrdict/61724/HTML/default/a000175758.htm Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Got the message Scott. Not being english my natural language, it is sometimes difficult for me to explain some matters with letters, it's way more simple to just code it 🙂 But I agree with you, and I'll make an effort to explain my suggestion rather than give them away. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Not sure about what your aiming to do, but, to stack KZ, you should use the SUBGROUP options. The following modification will suffice. proc gchart data=work.GrundlageGrafik; format datum ddmmyyp10.; vbar datum / sumvar = stapel group = gruppe subgroup = KZ /* datum */ legend = legend1 patternid = subgroup sum raxis = axis1 maxis = axis2 gaxis = axis3; axis1 label=none minor=none; axis2 label=none value=(tick=1 '' tick=2 ''); axis3 label=none order=('Bis 2 Tage' '3-5 Tage' '6-8 Tage' 'Aelter 8 Tage'); run; GROUP, will combine categories and SUBGROUP will stack values. Is this what you want? Check the online documentation for SAS Graph in particular for the gchart procedure: http://support.sas.com/documentation/onlinedoc/graph/index.html And give a look at the sample gallery: http://support.sas.com/sassamples/graphgallery/index.html Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Much more clearer now. OK, forget my previous suggestion. proc sort data = ADMISSIONS; by ID_PERSON DATE; run; /* sort */ data ADMISSIONS_C; set ADMISSIONS; by ID_PERSON; if first.ID_PERSON the COUNT=0; /* reset counter */ COUNT+1; /* count admission, implicitly retained */ run; Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

I'm sorry Susan. The syntax you posted is indeed correct, and will work as expected. Scott, the use of the variable _N_, is a common optimization technique which I believe firstly appeared by the hand of the great Paul Dorfman. _N_ is not used by SAS for counting rows, the row number is just passed to this automatic variable at the begin of each datastep loop. It's just meant to be used inside each loop, of course, if needed, which is not the case. So, if your not interested about its value, there's nothing wrong about changing it's value, and you get a perfectly good temporary variable, which no extra memory and that does not need to be dropped at the end of the datastep. Same goes for other automatic variable, like _ERROR_ and _IORC_. But this would not be the subject here, so returning to Susan's question, Scott's example is right, and will do what is expected. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

For the counting, a simple sql query will do the job. proc sql; select ID_PERSON, count(*) as COUNT from ADMISSIONS group by ID_PERSON; quit; Not sure about what you wish to do with the date ordering. Do you want to order the counts by the lastes admission date? Or do you wish to simply order the ADMISSIONS dataset? Cheers from Portugal. Daniel Santos @ www.cgd.pt Message was edited by: Daniel Santos ... View more

Hello Susan. the use of, if last.ID; is know as a subsetting if. only rows satisfying the if condition will pass beside the if (an reach the run; and so implicit output). since after that, there is a do loop, with explicit output, this loop will be executed only at the last ID row. Be careful that the syntax you have posted will not work, you should change it to: if last.ID then do; do _N_=1 to _C; output; end; and this will do exactly the same as my code, because the do loop will only be executed at the last ID row, also. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more

Have you provided a XML map for that data? It seems you have a complex structure in there. Cheers from Portugal. Daniel Santos @ www.cgd.pt ... View more