Date:         Sun, 30 Apr 2000 16:53:54 -0400
Reply-To:     WHITLOI1 <WHITLOI1@WESTAT.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         WHITLOI1 <WHITLOI1@WESTAT.COM>
Content-Type: text/plain; charset=US-ASCII

Subject: SAS V8 documentation on CD-ROM
Summary: A conflict in token lengths and an editor gotcha.
Respondent: Ian Whitlock <whitloi1@westat.com>

Thanks to

   Ray Pass <raypass@WORLDNET.ATT.NET>
   Dale McLerran <dmclerra@FHCRC.ORG>
   Puddin' Man <pudding_man@ALTAVISTA.COM>
   "Fehd; Ronald J." <rjf2@CDC.GOV>

for their support in pointing out problems with the V8 Online
documentation.  There is still plenty of room for more people to
join in.  As Ray points out I am not asking for serious study, just
look and point out when you use the product.

Here is one that hit my eye while perusing the SAS Macro Language
Reference.  Later I turned it into a search problem.

1) What is the maximum length of a token in the SAS system?

   From SAS Language Reference Concepts (System Concepts - Rules for
   Words and Names)

      >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
      A word or token in the SAS language is a collection of
      characters that communicates a meaning to SAS and is not
      divisible into smaller units capable of independent use. It
      can contain a maximum of 32,767 characters.
      <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

   From SAS Macro Language Reference (SAS Tokens)

      >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
      The maximum length of any type of token is 200 characters. A
      token ends when the tokenizer encounters one of the following
      situations:

         the beginning of a new token

         a blank after a name or number token

         in a literal token, a quote of the same type that started
         the token. There is an exception. A quotation mark followed
         by a quotation mark of the same type is interpreted as a
         single quotation mark that becomes part of the literal
         token. For example, in 'Mary''s', the fourth quotation mark
         terminates the literal token. The second and third
         quotation marks are interpreted as a single character which
         is included in the literal token.
      <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

   The macro guide needs updating to the current version.  Last
   December I posted a macro that depended on the word scanner
   recognizing tokens longer than 200 bytes.

2) Example from SAS Language Reference Dictionary - Systems Options

   Cardimage example:

      >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
      Consider the following DATA step:

      data;
         x='A B';
      run;

      If CARDIMAGE is in effect, the variable X receives a value
      that consists of 78 characters: the A, 76 blanks, and the B.
      If NOCARDIMAGE is in effect, the variable X receives a value
      that consists of two characters: AB, with no intervening
      blanks.
      <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

   Before the example there is this note:

      >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
      Operating Environment Information:   CARDIMAGE is generally
      used in the OS/390 and CMS operating environments; NOCARDIMAGE
      is used in other operating environments.
      <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

   What does one make of this?  I assumed, that the assignment for X
   should have been on two lines instead of one.  I then tried the
   following under *Windows 98*.

   13   options nocardimage ;
   14
   15   data wno ;
   16      x = 'a
   17      b' ;
   18      put "----+----+" / x $char80. ;
   19   run ;

   ----+----+
   a   b
   NOTE: The data set WORK.WNO has 1 observations and 1 variables.
   NOTE: DATA statement used:
         real time           0.16 seconds


   20
   21   options cardimage ;
   22
   23   data wyes ;
   24      x = 'a
   25      b' ;
   26      put "----+----+" / x $char80. ;
   27   run ;

   ----+----+
   a   b
   NOTE: The data set WORK.WYES has 1 observations and 1 variables.
   NOTE: DATA statement used:
         real time           0.05 seconds


In addition to the mistake in presentation of the code, there seems
to be something missing here.  Aha! The above code was submitted
using the Enhanced Editor.  When using the program editor the results
for CARDIMAGE were

   51   options cardimage ;
   52
   53   data wyes ;
   54      x = 'a
   55      b' ;
   56      put "----+----+" / x $char100. ;
   57   run ;

   ----+----+
   a                                                                          b
   NOTE: The data set WORK.WYES has 1 observations and 1 variables.
   NOTE: DATA statement used:
         real time           0.32 seconds

When I checked the length of X it was $76 not $78, but then we
really don't know what in what column "b" was placed.  I am more
disturbed by the discovery that the same code produces different
results depending on whether the program editor or the enhanced
editor was used.

For me, it is not a big problem.  I never have and do not intend to
take advantage of the CARDIMAGE option.  But it does make me wonder
what other surprises will come out of using the Enhanced editor.  In
general I like this editor, but still depend on TextPad for serious
program writing.  (The tab feature - submit only one line - threw me
for a loop when I started using V8.)

Ian Whitlock