Hi,

I tried to make a simple comparaison like this :

 
-IF '-' EQ '+' THEN GOTO :ADDFEES;  
-SET &VAR1 = 10;
-GOTO :ENDFEES
-:ADDFEES
-SET &VAR1 = 20;
-:ENDFEES
-TYPE VAR1 = &VAR1

And as a result I have :
VAR1 = 20

Does somebody know this bug?

PS: I tested with the 7.6.4 version and I become the same wrong result...

This message has been edited. Last edited by: Jérény Deidda,

Hello everybody,

Just try to put a \ character before the +

Dr Nick

DM variables are all stored as strings and their type is determined at runtime by parsing the data according to the operator.

Thus depending on the operator concerned
' 1234.56 ' can be a string or a double precision number

Example: the equality operator attempts to convert both sides to numbers where possible so

' 1234.56 ' EQ '1234.56' will be true.

This is the behaviour in your case. WebFOCUS assumes '+' and '-' are numbers (incorrectly) and converts both to zero.

To ensure an alphanumeric operand in any test concatenate a blank

eg instead of
-IF &A EQ &B THEN ...

do

-IF &A | ' ' EQ &B | ' ' THEN ...

Painful I know. In retrospect dm variables may have been better with types but that would make them less flexible but less confusing!

There is an easier and more robust way to do this kind of equations in DM.
Use the ASIS functions like:
-IF ASIS('-') EQ ASIS('+') THEN GOTO :ADDFEES;

Thx for all these explanations.

GamP

You're right ASIS would be the correct way to do this but it does not work in subroutine arguments when you want to force an alpha, that's why I stick to blank concat cos it always works.

Regards

John

Another bug:

 
-SET &TEST = 0-0.1;
-TYPE &TEST
 

I become 0 instead of -0.1. Why?

Integer minus decimal is integer...

I tried 0.1 - 0.2 and the result is the same.

Problem is resolved by a setting in 7.6. Prior releases have rounding of results to integer. If you need to retain the precision multiply by a power of 10.

Try 100 * (0.32 - 0.13) should be 19.

The problem is that I don't know the precision before making the subtraction. Another idea?

Furthermore, I tried this:

  
-SET &TEST = 19/100;
-TYPE &TEST 
-QUIT

The result is always WRONG : 0 instead of 0.19

This message has been edited. Last edited by: Jérény Deidda,

Jérény,

Amper variables have always been dealt with as alphanumeric or integer. So what you are experiencing is not a bug.

T

The best approach is to use FTOA:

-SET &TEST = FTOA(0.32 - 0.13, '(F8.2)', 'A8');
-TYPE &TEST 

-SET &TEST = FTOA(19/100, '(F8.2)', 'A8');
-TYPE &TEST 

until you get 7.6 and use the DMPRECISION setting.
As Tony says, not a bug, but expected behavior.

OK thanks for this solution.

I called IBI support and an employee told me that the best way is to avoid making operations with th SET command and to make that in a DEFINE FILE or in a TABLE FILE when it's possible.