Beware of operator precedence

[David Marso]

Well, I bumbled into something yesterday that I should have been able to fix immediately.
After a bit of fumbling it became face palm time.
Who woulda thunk?
It has to do with operator precedence.
DOH!!!!
---
DATA LIST FREE / x.
BEGIN DATA
1
END DATA.
LOOP #a=1 TO 2.
LOOP #b=1 TO 2.
COMPUTE a=#a.
COMPUTE b=#b.
COMPUTE aneg0  =   -1**#a.
COMPUTE bneg0  =   -1**#a.
COMPUTE abneg0 = aneg0*bneg0.
COMPUTE aneg1  = (-1)**#a.
COMPUTE bneg1  = (-1)**#b.
COMPUTE abneg1 = aneg1*bneg1.
XSAVE OUTFILE 'ab.sav' / KEEP a b aneg0 bneg0 abneg0 aneg1 bneg1 abneg1.
END LOOP.
END LOOP.
EXE.
GET FILE 'ab.sav'.
LIST.

       A        B    ANEG0    BNEG0   ABNEG0    ANEG1    BNEG1   ABNEG1

    1.00     1.00    -1.00    -1.00     1.00    -1.00    -1.00     1.00
    1.00     2.00    -1.00    -1.00     1.00    -1.00     1.00    -1.00
    2.00     1.00    -1.00    -1.00     1.00     1.00    -1.00    -1.00
    2.00     2.00    -1.00    -1.00     1.00     1.00     1.00     1.00


Number of cases read:  4    Number of cases listed:  4

[Richard Ristow]

At 09:20 AM 3/28/2013, David Marso wrote:

>Well, I bumbled into something yesterday ... It has to do with
>operator precedence.
>---
>LOOP #a=1 TO 2.
>LOOP #b=1 TO 2.
>COMPUTE a=#a.
>COMPUTE b=#b.
>COMPUTE aneg0  =   -1**#a   /* Always returns -1 */.
>COMPUTE bneg0  =   -1**#a   /* Always returns -1 */.
>COMPUTE abneg0 = aneg0*bneg0.
>COMPUTE aneg1  = (-1)**#a   /* Works as desired  */.
>COMPUTE bneg1  = (-1)**#b   /* Works as desired  */.

Fascinating. So exponentiation takes precedence over monadic "-". I
wonder whether the precedence rules are documented anywhere?  The
word "precedence" doesn't occur in the Universals section of the
Command Syntax Reference.

Monadic, or prefix, "-" is subject to a certain ambiguity when it
precedes a number: Is is a part of the constant, and so has
precedence over everything, or is it an operator? And, if an
operator, with what precedence? I notice that C++ gives it very high
precedence, so your code would have worked. I wonder what language or
recommendations SPSS was following, developing its precedences rules?
They must date from the 1970s.

By the way, David, since you're a lover and master of the neat coding
technique, here's how to run your test code and demo without DATA
LIST or XSAVE:

NEW FILE.
INPUT PROGRAM.
LOOP #a=1 TO 2.
LOOP #b=1 TO 2.
COMPUTE a=#a.
COMPUTE b=#b.
COMPUTE aneg0  =   -1**#a.
COMPUTE bneg0  =   -1**#a.
COMPUTE abneg0 = aneg0*bneg0.
COMPUTE aneg1  = (-1)**#a.
COMPUTE bneg1  = (-1)**#b.
COMPUTE abneg1 = aneg1*bneg1.
END CASE.
END LOOP.
END LOOP.
END FILE.
END INPUT PROGRAM.

LIST.
List
|-----------------------------|---------------------------|
|Output Created               |30-MAR-2013 12:25:03       |
|-----------------------------|---------------------------|
        a        b    aneg0    bneg0   abneg0    aneg1    bneg1   abneg1

     1.00     1.00    -1.00    -1.00     1.00    -1.00    -1.00     1.00
     1.00     2.00    -1.00    -1.00     1.00    -1.00     1.00    -1.00
     2.00     1.00    -1.00    -1.00     1.00     1.00    -1.00    -1.00
     2.00     2.00    -1.00    -1.00     1.00     1.00     1.00     1.00

Number of cases read:  4    Number of cases listed:  4

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Jon K Peck]

From the CSR in the Universals>Transformation Expressions section

The order of execution is as follows: functions; exponentiation; multiplication, division,
and unary -; and addition and subtraction.
�� Operators at the same level are executed from left to right.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Richard Ristow <[hidden email]>
To:        [hidden email],
Date:        03/30/2013 10:50 AM
Subject:        Re: [SPSSX-L] Beware of operator precedence
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

At 09:20 AM 3/28/2013, David Marso wrote:

>Well, I bumbled into something yesterday ... It has to do with
>operator precedence.
>---
>LOOP #a=1 TO 2.
>LOOP #b=1 TO 2.
>COMPUTE a=#a.
>COMPUTE b=#b.
>COMPUTE aneg0  =   -1**#a   /* Always returns -1 */.
>COMPUTE bneg0  =   -1**#a   /* Always returns -1 */.
>COMPUTE abneg0 = aneg0*bneg0.
>COMPUTE aneg1  = (-1)**#a   /* Works as desired  */.
>COMPUTE bneg1  = (-1)**#b   /* Works as desired  */.

Fascinating. So exponentiation takes precedence over monadic "-". I
wonder whether the precedence rules are documented anywhere?  The
word "precedence" doesn't occur in the Universals section of the
Command Syntax Reference.

Monadic, or prefix, "-" is subject to a certain ambiguity when it
precedes a number: Is is a part of the constant, and so has
precedence over everything, or is it an operator? And, if an
operator, with what precedence? I notice that C++ gives it very high
precedence, so your code would have worked. I wonder what language or
recommendations SPSS was following, developing its precedences rules?
They must date from the 1970s.

By the way, David, since you're a lover and master of the neat coding
technique, here's how to run your test code and demo without DATA
LIST or XSAVE:

NEW FILE.
INPUT PROGRAM.
LOOP #a=1 TO 2.
LOOP #b=1 TO 2.
COMPUTE a=#a.
COMPUTE b=#b.
COMPUTE aneg0  =   -1**#a.
COMPUTE bneg0  =   -1**#a.
COMPUTE abneg0 = aneg0*bneg0.
COMPUTE aneg1  = (-1)**#a.
COMPUTE bneg1  = (-1)**#b.
COMPUTE abneg1 = aneg1*bneg1.
END CASE.
END LOOP.
END LOOP.
END FILE.
END INPUT PROGRAM.

LIST.
List
|-----------------------------|---------------------------|
|Output Created               |30-MAR-2013 12:25:03       |
|-----------------------------|---------------------------|
       a        b    aneg0    bneg0   abneg0    aneg1    bneg1   abneg1

    1.00     1.00    -1.00    -1.00     1.00    -1.00    -1.00     1.00
    1.00     2.00    -1.00    -1.00     1.00    -1.00     1.00    -1.00
    2.00     1.00    -1.00    -1.00     1.00     1.00    -1.00    -1.00
    2.00     2.00    -1.00    -1.00     1.00     1.00     1.00     1.00

Number of cases read:  4    Number of cases listed:  4

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Richard Ristow]

At 01:24 PM 3/30/2013, Jon K Peck wrote:
> From the CSR in the Universals>Transformation Expressions section
>
>The order of execution is as follows: functions; exponentiation;
>multiplication, division, and unary -;
>and addition and subtraction.
>Operators at the same level are executed from left to right.

Thank you, Jon. And my apologies, for having searched superficially
for this information.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD