expr problem 1

[marsd]

Hi all.

I have a procedure that acts as a calculator.
For multiplication ,addition and subtraction
it works well...The guts of the procedure is:

for {set r $num} {$r < [llength $sumlist]} {incr r} {
set sum [expr $sum == 0 ? [lindex $sumlist $r] : [expr $sum * [lindex $sumlist $r]]]
r_calc $sumlist $op $sum [expr $r + 1]
}

#main()
set a(1) [r_calc {12 54 67 89 2} / 0 0]
set a(2) [r_calc {32 45 12} + 0 0]
set a(3) [r_calc {43 56 78} - 0 0]

b_parray a
#end

The output of this sample:
a(1) = 2
a(2) = 89
a(3) = -91

a(1) should be something like 1.86334e-05 according
to a calculator.

Anybody know what the issue is?

 

[AviaTraining]

Well, I'm having a bit of difficulty following the evaluation of your code because I think you truncated it a bit too much. But my suspicion, based on your description of the problem, is that you're tripping over expr's preference for doing integer math whenever possible. If both operands are integers, expr performs integer math; if at least one is floating-point, expr performs floating-point math. Some examples:

[tt]% expr { 4 / 3 }
1
% expr { 4.0 / 3 }
1.33333333333
% set x 7; set y 3
3
% expr { $x / $y }
2[/tt]

The best way to force expr to perform floating-point math is to use its built-in double() function on one of the operands to force its interpretation as a floating-point value. For example:

[tt]% expr { double(4) / 3 }
1.33333333333
% expr { double($x) / $y }
2.33333333333[/tt]

Hope this helps! - Ken Jones, President, ken@avia-training.com
Avia Training and Consulting,

http://www.avia-training.com

866-TCL-HELP (866-825-4357) US Toll free
415-643-8692 Voice
415-643-8697 Fax

 

[marsd]

Yeah, that was the whole problem.
I'd previously used double, but in this way:
set sum [expr $sum == 0 ? [lindex $sumlist $r] : [expr double($sum / [lindex $sumlist $r])]], and had the same problem. Using your example it
works.
Thanks

 

[AviaTraining]

Yes, it's a very common mistake to make. In case there are others reading this thread and not seeing the problem, double() didn't work the way marsd first tried it because he put it around the entire expression:

[tt]expr double([ignore]$sum / [lindex $sumlist $r][/ignore])[/tt]

So, expr evaluated what was inside the parentheses first -- doing the (integer) division -- and then converting the result of the division to floating-point. - Ken Jones, President, ken@avia-training.com
Avia Training and Consulting,

http://www.avia-training.com

866-TCL-HELP (866-825-4357) US Toll free
415-643-8692 Voice
415-643-8697 Fax