Date: Mon, 13 Aug 2001 10:13:46 -0700
Reply-To: Dale McLerran <dmclerra@MY-DEJA.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Dale McLerran <dmclerra@MY-DEJA.COM>
Subject: Re: Interesting difference between 6.12 and 8.2
Content-Type: text/plain
Greg,
One of the big changes between version 6.xx to 8.y was support for
long variable names. At the same time that they implemented support
for long names, SI also adopted mixed case naming. This support for
mixed case naming is not like Unix mixed case naming, where TEST,
Test, and test specify different variables. In SAS, there can be
only one variable named TesT, no matter how you choose to write it
employing upper and lower casing. Moreover, the name is stored with
the particular casing as the first occurrence of the variable. So,
if you had the statements
data myv8file;
length TEST 3;
test = 1;
output;
run;
then the single variable in this dataset would be placed in the
descriptor as TEST. However, if you reversed the order of the length
and assignment statements in the above code, the variable would be
place in the descriptor as test.
What I have described so far is the default behavior in version 8.
Now, realizing that this could cause some folks problems, SAS has
implemented a system option which allows you more control over the
naming conventions. First of all, you can adopt a version 6
convention, in which names have no more than 8 characters and are
stored as uppercase. Alternatively, you can allow names to have
length up to 32 characters but require that the names be stored in
uppercase, regardless of how the name first appears. The system
option which controls how SAS names are stored is the option
VALIDVARNAME. It can take on values V7 (default), V6 (max 8 bytes
uppercased), UPCASE (max 32 bytes, uppercased), or ANY. Note that
if you specify ANY, SAS can employ variable names which have any ANSII
characters. Thus, you could have variable names which have embedded
spaces, names which use parentheses, etc.
I have found this most useful when reading Access tables, where the
column names are often not "variable names" but rather "variable
descriptions". In order to reference variables with embedded spaces
or other characters which are not part of the traditional SAS naming
convention, SAS has implemented what are called name literals. Name
literals have the variable name surrounded by quotation marks,
followed by the letter n. Thus, you could reference a name literal as
"Sodium Chloride"n
I know that this is the long answer to your short question, but
understanding the available naming conventions will help you get the
most out of version 8.
Dale
>Date: Mon, 13 Aug 2001 10:49:21 -0500
>Reply-To: greg.woolridge@TAP.COM
> Greg Woolridge <greg.woolridge@TAP.COM> Interesting difference between 6.12 and 8.2 SAS-L@LISTSERV.UGA.EDU
>Hi SAS-Lers,
>
>We are finally getting round to investigating 8.2 and how converting from
>6.12 will affect us. I have noticed something interesting with data sets
>in 8.2 vs 6.12 an was wondering if anyone had run into and dealt with this.
>
>The following bit of code produces identical results in 6.12, but not in
>8.2:
>
>data tester;
> input @1 comment $35. @37 comm1 $35.;
> cards;
>This is a comment to test mixed This is the second comment variable
>THIS IS ANOTHER COMMENT and this is lower case
>This contains SOME UPPER and lower This one IS ALSO upper and LOWER
>let's try one that is all lower AND ONE THAT IS ALL UPPER
>This Is The Last One This is the last one
>;
>run;
>
>proc contents; run;
>
>data tester;
> input @1 COMMENT $35. @37 comm1 $35.;
> cards;
>This is a comment to test mixed This is the second comment variable
>THIS IS ANOTHER COMMENT and this is lower case
>This contains SOME UPPER and lower This one IS ALSO upper and LOWER
>let's try one that is all lower AND ONE THAT IS ALL UPPER
>This Is The Last One This is the last one
>;
>run;
>
>proc contents; run;
>
>If you look closely, you will notice that the difference between the 2 data
>steps is that the input variable comment in the second step is in all caps.
>In 6.12 the contents procedure shows all variables in all caps, regardless
>of how they're typed in the program code. Hence, the result is the same in
>both cases. In 8.2, the contents procedure shows the variable names in the
>same case as is used in the program . The 2 steps give different results.
>I have discovered that the effect is apparently in the descriptor portion
>of the data set, as the differences are also noticed in the options window
>of Display Manager and when using other SAS procedures and functions to
>compare variable names.
>
>I can't find any documentation on this change or any option that will allow
>me to go back to the old way. The problem is that I have some programs
>where I do a PROC CONTENTS and use a statement like if name='CHECK'. Now
>it appears that I will need to change all that code to something like if
>upcase(name)='CHECK'.
>
>Has anyone else run into this? Can anyone offer a better solution than
>recoding all programs?
>
>Greg M. Woolridge
>Manager, Study Programming
>TAP Pharmaceutical Products Inc.
>e-mail: greg.woolridge@tap.com
>phone: 847-582-2332
>fax: 847-582-2403
---------------------------------------
Dale McLerran
Fred Hutchinson Cancer Research Center
mailto: dmclerra@fhcrc.org
Ph: (206) 667-2926
Fax: (206) 667-5977
---------------------------------------
------------------------------------------------------------
--== Sent via Deja.com ==--
http://www.deja.com/
| 
