Date: Tue, 8 Jul 2003 16:31:06 -0700
Reply-To: "William W. Viergever" <wwvierg@ATTGLOBAL.NET>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "William W. Viergever" <wwvierg@ATTGLOBAL.NET>
Subject: Re: PROC SUMMARY Failure
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 06:34 PM 07/08/2003 -0400, Sigurd Hermansen wrote:
>The 'valid output statement' phrase differs from anything that I can recall
>seeing. I have not, however, tried to overwrite the source dataset with the
>output of PROC SUMMARY. Perhaps SAS V8.2 is trying to protect less wary
>users against loss of data.
Sorry for earlier glib repsonse - it was well intended as I've seen that
msg myself (but alas, can't remember when/why).
But after a very parsimonious <g> private reply from Sig, I tried some code
here and found I was able to over-write the input dataset.
Bottom-line: no answer here; sorry.
>You may have to resort to a SAS SQL query as a workaround. If you ask SAS
>SQL to shoot you in the foot, it will gladly optimize the process. ;>
>From: Kevin Myers [mailto:WHMyers@CABLEONE.NET]
>Sent: Tuesday, July 08, 2003 4:43 PM
>Subject: PROC SUMMARY Failure
>As some of you may recall, using PROC SUMMARY in the following way used to
>work up through early V8.2 at least. Note that there is no VAR statement,
>and the input data set has no numeric variables. I have used this method
>many times over the years in order to obtain a simple list of all
>combinations of class variable values:
>proc summary data=Country missing nway order=freq;
> class IHSCountryCode DSCountryId CountryName CountryAbbr;
> output out=Country;
>However, I now receive the following error message (SAS 8.2 with all?
>hotfixes under Win 2K):
>ERROR: Neither the PRINT option nor a valid output statement has been given.
>This or some similar message used to be produced under equivalent
>circumstances when using PROC MEANS, but NOT when using PROC SUMMARY
>(despite what the docs might lead you to believe to the contrary). Has
>someone changed this PROC SUMMARY behavior on purpose??? If so, this is
>going to break LOTS of existing programs!!!
>If anyone can see some stupid mistake that I'm making, or has any
>information regarding this apparant change in behavior, I'd really
>appreciate it. Thanks!