C compiler and BOOL_WIDTH

Semphalk — Fri, 09 Feb 2024 16:26:00 GMT

I am using:

Intel(R) oneAPI DPC++/C++ Compiler for applications running on Intel(R) 64, Version 2024.0.2 Build 20231213

I am running:

icx -Qstd=c2x myfile.c

The standard C23 (still in drat mode, we have to say),

introduces the macro BOOL_WIDTH in <limits.h>.

As far I understand, the suffix _WIDTH for every constant appearing

in ISO standards of C language

refers to the number of "value bits" of the corresponding data type.

(a) The compiler reports a value of BOOL_WIDTH == 8.

(b) On the other hand, sizeof(bool) == 1.

By the conditions (a) and (b) I conclude that there is not room for trap representation.

Thus, there are 256 effective choices of bit values to represent the values 0 and 1.

In this way, the type 'bool' is using many different internal representations (in terms of bits) in order to store the same mathemical value.

My doubt here is which combinations of bits should represent the boolean value 'zero' and which combinations of bits are used to represent the value 'one'.

I am guessing that every combination of bits containing at least one bit flagged to 1 is used to represent the boolean value 'one' and that 'zero' is represented only with the combination of bits containing all bits set to 0.

But I don't know if such a behaviour is well documented.

Re: C compiler and BOOL_WIDTH

Alex_Y_Intel — Mon, 08 Apr 2024 19:19:34 GMT

Hi @Semphalk can you please provide myfile.c?

topic C compiler and BOOL_WIDTH in Intel® oneAPI DPC++/C++ Compiler

C compiler and BOOL_WIDTH

Re: C compiler and BOOL_WIDTH