When PROC FedSQL executes code on the SAS Compute Server (what you might call base SAS), it can ONLY access data from non-concatenated SAS libraries which are not associated with a caslib. When executing in CAS (as indicated by the presence of the SESSREF= option on the PROC FedSQL statement) it can ONLY access data in caslibs.  When you are operating in a multi-user CAS session, every user has a caslib named CASUSER. In order to differentiate your CASUSER caslib from other users, the system identifies the caslib as CASUSER(userID).  So you need to use the actual CASLIB name for this, not the libref. Of course, the caslib name will seem inappropriate to the FedSQL compiler because of the included punctuation.  To make the name work as is, you can add double quotes to indicate that the text is a valid name, something like this: proc fedsql SESSREF=&_SESSREF_; create table "&mp_sourceName"."tmp_&inputTable" as select a.* ,b.time_id as start_dt_sk from "&mp_sourceName"."&inputTable" a "&mp_sourceName"."ed_&inputTable" b where a.%bquote(&time_dim.) between b.start_date and b.end_date ; quit; I hope this helps. Mark ... View more

The reason this doesn't work is given in the log: error: ERROR: InventoryFulfillment is not a valid SAS name. A caslib name can be up to 256 characters long, but a SAS libref has a maximum length of 8 characters. So InventoryFulfillment is valid as a caslib name but is not valid as a libref.  You still have not shared your code so more specific help is not possible, but when you assign the libref to the caslib, you will need to supply a libref value that meets the 8-character maximum criteria.      ... View more

Thanks ... View more

Had you not told me, I wouldn't know.  Odpowiednio, dziękuję bardzo!  ... View more

issue resolved, I was using wrong table reference . thank you ... View more

@Aexor wrote: Sorry. I need one more info. Cas we have CAST kind of functionality outside pass through query. To use CAST function in later step . I have to load the dataset in PG schema and then I am using CAST . Can we remove the LOADING overhead and use CAST like functional in PROC SQL itself . Thank You! Are you talking about loading the data first into SAS and then do the join within SAS? If so then SAS 9.4 got only two data types (character and numeric). A UUID column will get converted to a character type column same as a Varchar so you wouldn't need any type casting anymore. https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/acreldb/p1prjok4ko5rf4n1h2rf5ok8d2ct.htm    With SAS type casting is done via put/input functions.   ... View more

A macro variable is already part of SAS. How does the data initially arrive in your SAS environment? And if the macro variable is filled from data out of a dataset, you can use the same step to create the code (statements are not limited in length), or use that extract in joins/hashes. ... View more

@Aexor wrote: Thank you all for your response. I can not post my log here, due to privacy issue. I can't copy those datasets in my local .   Create data sets with different variable names and non-sensitive information that behaves like your real data. Demonstrate the use with the dummy data. Make sure you can determine the desired result from your dummy data and show that.   ... View more

What data connector do you use? Can you access the database from SAS compute (e.g., from SAS Studio via a LIBANME statement)? What have you tried, please attach a log. ... View more

Normal SAS is probably smart enough to do that.  What is does when you are creating a NEW dataset like that is write the dataset first.  Then when it finishes writing successfully it deletes the old one and renames what it just wrote to the desired name.   But if &LBNAME is a libref that is defined using SAS/Access to some external database then you have to worry about what that database allows.  Most will not let you create a dataset with the same name as an existing dataset. ... View more

You have not explained what "copying step causing issue" is. Please post the SAS log of the copying step including any notes and errors. ... View more

This type of error can happen if you copy and paste the code from a browser or formatted document to a text file. ... View more

Just in case the actual data has more than one year in the data set I might suggest use of the YYMON7. format . That will show column years such as 2012JAN -2012DEC, 2013JAN -2013DEC. That way you don't have to do any code for multiple year datasets.   Unless you need that output as a data set for further manipulation/analysis I would use a report proc such as Tabulate or report data have; input product $ date :date9. sales; format date date9.; datalines; a 1Jan2012 200 b 4Jan2012 300 c 2Feb2012 200 d 3Mar2012 200 d 5Mar2012 100 ; proc tabulate data=have; class product date; format date yymon7.; var sales; tables product, date=' '*sales*sum=' '*f=best10. ; run; proc report data=have; columns product date, sales; define product/group; define date/across format=yymon7. order=data; run; ... View more

Yes. Done that. Just one more information . I have used this code and this also worked fine data have; input id name $ lastname $ number; infile datalines dlm = '|'; datalines; 1|John| |. 1| |Deo|. 1| | |8888 2|John| |. 2| |Deo|. 2| | |8888 ; run; data want; update have(obs=0) have; by id; /* if last.ID then output; */ run; Could you please tell me significance of last.id in your code. ... View more

You can do this with SQL or a data step.   SQL example: proc sql; update lib.data1 set a=(select a2 from lib.data2 where id=data1.id) where id in(select id from lib.data2 where salesdate>'01JAN2020'd) ; quit; Or the same thing in a data step: data lib.data1; set lib.data2; where salesdate>'01JAN2020'd; do until(0); modify lib.data1 key=id; if _iorc_ then leave; a=a2; replace; end; _error_=0; run; The data step example as shown needs LIB.DATA1 to be indexed by ID.  ... View more