VI of the Day (10/27/2009) - Boolean to (0,1)

[Darin.K]

Sticking with the efficiency theme, this one has me (and a few others) scratching their heads, "Boolean to (0,1)".  Seems pretty straightforward, boolean in and either a 0 or 1 out.  What is interesting is the choice of I16 for the representation of 0 or 1.  It isn't so much that you are now using 16-bits to hold 1 bit of information, but how often do you use the I16 data type?  If you are indexing an array or loop you are using I32 (or coercing), if you are working with byte arrays you use U8.  Which now makes me think, why the heck do you use a signed representation for 0/1?

 

My guess is that in the old days booleans were stored in two bytes (it must have seemed like a good idea at the time), and this is a remnant.  What we are left with is a potentially useful function that we have to either follow with a conversion, replace with Select or Type Cast or live with a coercion.

I believe there is a movement afoot in the idea exchange to get this changed. 

VIOTD groundrules here. 

[smercurio_fc]

Actually, Booleans are stored as 8-bit values. It used to be that they were stored as 1-bit values. But that was a loooooong time ago. This little tidbit is what makes this little trick possible.

[PaulG.]

Could never figure out why NI made this output I16. Your explanation seems reasonable.

[smercurio_fc]

Hmmm... You know, now that I think about it you may be right about the 16-bit storage. I could swear, though, there was a 1-bit storage of something at one point. Of course, I may be confusing it with another language.

[Darin.K]

---

smercurio_fc wrote:
Actually, Booleans are stored as 8-bit values. It used to be that they were stored as 1-bit values. But that was a loooooong time ago. This little tidbit is what makes this little trick possible.

---

 

Booleans are now one byte in or out of arrays.  Used to be they were 2 bytes outside of an array and 1-bit inside of an array.  The 1-bit representation is still very useful, especially when dealing with image masks.  The one byte representation is very useful for code obfuscation, if nothing else.  Is there anything else?   

[blawson]

I use this guy when reading an error cluster or other boolean to a 3-way LED.  Not sure if this is the best approach, but ?1:0 saves me a Select and some wiring.  The 3way LED is a 3 item pict ring with images for off, green, and red.

(LV2009f2)

[smercurio_fc]

---

Darin.K wrote:
Booleans are now one byte in or out of arrays.  Used to be they were 2 bytes outside of an array and 1-bit inside of an array. 

---

Ah-hah! That was it. I knew there was a 1-bit storage of something in the dark, dreaded past. Your memory is clearly better than mine.  

[RandyP]

---

smercurio_fc wrote:
This little tidbit is what makes this little trick possible.

---

I can't figure out how that works. It actually seems pretty scary that others were able to do the same thing, but with other strings coming out. Can anyone explain that to me?

[GerdW]

Hi Randy,

two facts:

1) a boolean is (nowadays) stored as 8bit entity internally

2) any value different than zero is TRUE

Do you need more hints?

[Ben]

---

Darin.K wrote:

Sticking with the efficiency theme, this one has me (and a few others) scratching their heads, "Boolean to (0,1)".  Seems pretty straightforward, boolean in and either a 0 or 1 out.  What is interesting is the choice of I16 for the representation of 0 or 1.  It isn't so much that you are now using 16-bits to hold 1 bit of information, but how often do you use the I16 data type?  If you are indexing an array or loop you are using I32 (or coercing), if you are working with byte arrays you use U8.  Which now makes me think, why the heck do you use a signed representation for 0/1?

 

My guess is that in the old days booleans were stored in two bytes (it must have seemed like a good idea at the time), and this is a remnant.  What we are left with is a potentially useful function that we have to either follow with a conversion, replace with Select or Type Cast or live with a coercion.

I believe there is a movement afoot in the idea exchange to get this changed. 

VIOTD groundrules here. 

---

The real reason can only be provided by someone who was in the meeting can tell us for sure but I'll speculate...

Since LV is cross-platform capable, ther various branching operations supported by the mulitude of patforms LV was targeted at was taken into concideration. A common machine operation is to "branch if negative". By writing all bits as true or false when a branch if negative is executed the MSB of the I16 is set making it negative.

So I suspect the "lowest common denomenator" across platforms was a branch if negative.

Rolf (if he is listening) is much more familair with the platform abstraction layer than I so maybe he could shed more than speculative light on this Q.

Ben