Date: Fri, 20 Dec 2002 10:37:13 +0800
Reply-To: =?gb2312?q?Johnson=20Chang?= <jchangmail@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: =?gb2312?q?Johnson=20Chang?= <jchangmail@YAHOO.COM>
Subject: Re: words on reports changed to lower case?
Content-Type: text/plain; charset=gb2312
Is there a standard procedure to migration with different version? are there some
comprehasive tech reports talking about these migration?
if differences are found, how to settle them? i means not all differences are resulted in
the same change. e.g i found the normality test output different values from v6 to v8,
and some "out= " statements in proc behavour differently, so the created datasets are
somewhat different as you mentioned.
How to process this difference when we finalized the project report? it is also a problem
to me these days when we almost finished a study.
Any suggestions, comments or references are appreciated,
--- Lex jansen <l.jansen@LEX-JANSEN.DEMON.NL> 的正文：> Sterling,
> At the pharmaceutical company I work for, we did a 'SASv6 to SASv8
> migration' validation.
> We took 16 completed trials, each 30-40 SAS programs and ran all those
> programs in SASv8 again.
> Then we compared the datasets that were created and we compared the output
> (listings, tables, analyses) that was created.
> For comparing the datasets we used a SAS macro that used PROC COMPARE.
> Some differences would show up because we had the creation date in dataset
> Also we noticed that the ROUND function had changed ...
> Our listings, tables and analyses have the creation date as well in a
> For comparing ASCII output we used the CSDIFF utility
> Before we compared, we stripped out the date dependent footnotes.
> Most trouble we had was because SASv8 changed it's behaviour with respect to
> ORDER variables in PROC REPORT.
> (SASNOTE: SN-003866: DISPLAY variable order in the COLUMN statement has no
> effect on the collating sequence of GROUP and ORDER variables)
> Hope this helps,
> Lex Jansen
Do You Yahoo!?