On Wed, 7 Feb 2007 20:11:23 -0000, Peter Crawford <Peter.Crawford@BLUEYONDER.CO.UK> wrote:

>I'm completely in agreement with Mr _null_

We have a profound gender imbalance in this group, so until more information surfaces, can we assume it's Ms. Null?

> >My posting was made to introduce the logical alternative to sql into :mVar >that is relevant when the information flows from processing in a data step. > >It needs to be appropriate, rather than large. I have no need or wish to >accumulate information from many rows in a single variable for the sole >purpose of loading a macro variable. > >Peter Crawford >Crawford Software Consultancy >UK > > >-----Original Message----- >From: data _null_; [mailto:datanull@gmail.com] >Sent: 07 February 2007 14:08 >To: Peter Crawford >Cc: SAS-L@listserv.uga.edu >Subject: Re: Creating 1 macro variable for multiple rows in dataset > >No you don't have to worry about the 32K limit rather the 64K limit. > >5851 + ,"Judy" >ERROR: The text expression length (65538) exceeds maximum length >(65534). The text expression has > been truncated to 65534 characters. > > >Instead of figuring out ways to put more data into macro variables we >should be thinking about ways to keep data out of macro variables. > >We (the L) are always warning users about how bad it is to put data >into meta data but we are happy to show folks how to put more data in >more and more macro variables. > >I've seen code to put an entire data set into macro variables. While >I never say never there is surely a better way to accomplish whatever >the task may be, than taking data out of a data set and putting it >into macro variables. > >I asked the OP to tell us more about what he is trying to accomplish >but he has not seed fit to provide that information. > >I suspect that if the OP told us what he wanted to do with these >strings then there would be many "better" solutions that do not use >macro variables at all, and are more efficient. > >SAS has many more powerful tools than MACRO. The data step for one. > > >On 2/7/07, Peter Crawford <peter.crawford@blueyonder.co.uk> wrote: >> another datastep approach ~ which won't ru out of the 32K limit >> of data step variable widths >> >> This method uses call execute() to stack a %let statement; >> >> %let sourceD = sashelp.class; ** a simple demo dataset; >> %let sourceV = name ; >> >> %let target= ; *initialize your receiving macro variable; >> data _null_; >> call execute( '%nrstr(%%)let target= ' ); >> do until( end_of_source ); >> set &sourceD end= end_of_source ; >> call execute( ',' !! quote( trim( &sourceV )) ) ; >> *on principle, I always use function quote() >> to deal with any embedded quotes correctly ; >> end; >> call execute( ';' ); >> stop; >> run; >> %let target = %substr( %superq(target), 2 ) ; >> *** removing initial comma; >> >> %let target = %sysfunc( translate( %superq(target) >> ,%str(%'),%str(%") ) ); >> *** replacing double- with single-quotes; >> >> %let target = %sysfunc( tranwrd( %superq(target) >> ,%str( ,),%str(,) ) ); >> *** removing 3 blanks preceeding each comma >> >> This has all been tested on windows-XP-pro. >> Your platform may implement differently, so beware!; >> ; >> >> Peter Crawford >> >> >> >> On Tue, 6 Feb 2007 15:34:10 -0500, Fehd, Ronald J. (CDC/CCHIS/NCPHI) >> <rjf2@CDC.GOV> wrote: >> >> >> From: Ross, Michael D >> >> I'm looking to create 1 macro variable containing "y" (rows >> >> 1-3), with single quotes around each value and commas >> >> seperating them for each instance of "z". So, I would have 2 >> >> macro variable in the example below that would look like: >> >> >> >> &var1 = 'bbbbbbbb','dddddddd','ffffffff'; >> >> &var2 = 'hhhhhhhh','iiiiiiii'; >> >> >> >> Also, I know how to do this in Proc Sql - I was just >> >> wondering how to do it in the Data step. >> > >> >stick with sql; >> >you'll run out of space for your character variable >> >doing this in a data step >> > >> >DATA _Null_; >> >attrib Concatenation length = $ 200; >> >retain Concatenation ' '; >> >do until(EndoFile); >> > set Test end = EndoFile; >> > Concatenation = cat(Concatenation,y); >> > end; >> >call sumputX('var1',Concatenation); >> >stop; >> >run; >> >%Put Var1<&Var1.>; >> > >> >Ron Fehd the macro maven CDC Atlanta GA USA RJF2 at cdc dot gov >> > >> >--> cheerful provider of UNTESTED SAS code from the Clue?Gee!Wrx <-- >> > >> >> data test; >> >> input y $8. z $8. ; >> >> datalines; >> >> bbbbbbbb xxxxxxxx >> >> dddddddd xxxxxxxx >> >> ffffffff xxxxxxxx >> >> hhhhhhhh yyyyyyyy >> >> jjjjjjjj yyyyyyyy >> >> ; >> >> run; >> >> >> >> >>