Error in bit manipulation

[DrLock]

This seems like a math script error, but maybe I am missing something: The bit manipulation functions seem to handle 0 incorrectly:

>>bitshift(0,1)
ans =

           2    

I expected 0.

>>bitget(0, 1:3)
ans =

           1      0      0    

I expected 0 0 0.

Is there a fix for this or something I am missing?

Thanks

[altenbach]

What is your LaVIEW version?

I just tested in 8.5 and the bitget(0, 1:3) now correctly returns 0 0 0.

The bitshift function is still incorrect when the first argument is zero.

[DrLock]

Oops sorry forgot to list that ... I am running LabView 8.2

[SimonH]

DrLock, altenbach,

The help file entry for bitshift does specify that the first argument should be a "positive integer" -- making the output for any other input undefined.  However I agree with you both that this is not the way it should behave and have sent it to the developers for them to look at.  This was reported to R&D (ID# 4ELI7NVQ) for further investigation.

One possible workaround would be to simply check for a zero input with a snippet of code such as:

if a==0
    ans=0
else
    ans=bitshift(a,1)
end

I hope this helps.

Regards,

Simon H
Applications Engineer
National Instruments
http://www.ni.com/support/

---

from the LabVIEW Help

d = bitshift(a, b, c)

Description

Performs a bitwise shift on the input elements. For example, if a = 9 (1001), bitshift(a, 1) = 18 (10010) and bitshift(a, 1, 4) = 2 (0010).

Examples

Inputs

Name	Description
a	Specifies a scalar, vector, or matrix of positive integers. All elements of a must be less than bitmax.
b	Determines the shift sizes. Shifting a by b bits is equivalent to multiplying a by 2^b and then rounding to the nearest integer. b is a scalar, vector, or matrix of integers. b is the same size as a unless a or b is a scalar.
c	Specifies the number of valid bits for each of the shifted integers. If the shifted integers exceed c bits, LabVIEW ignores the overflow. c is a matrix of integers. The default is 53.

---

[altenbach]

Hmm... so the labview help is more restrictive here for no good reason.

I was consulting the matlab help, which is more general:

"C = bitshift(A, k) returns the value of A shifted by k bits. Input argument A must be an unsigned integer or an array of unsigned integers."

Clearly, the function should be defined the same in both environments. 🙂