When to use WordBool and LongBool?

[Handoko]

The size of Boolean also ByteBool is 1 byte only. Why do we need the bigger WordBool and LongBool? What are the use cases of them?

[balazsszekely]

It is most useful when interfacing with C and C++ application. Under windows I used extensively on winapi calls.

[Thaddy]

Yes. It is the relation to other languages, but not only that.
In computer science - you know that - only false is defined enough to be usable, it being zero, empty. True is simply not false and that is undefined, but unequal to zero.
How a language implements that is really... ahum...language dependent. Hence we have multiple options in Pascal to describe boolean values.

To complicate matters further, since 3.0 we can also redefine true/false as e.g. string. Or pointers...

Code: Pascal  [Select][+][-]

1. const false=nil;

but that gives you a nice "feature" that you can't possibly define true...At least I couldn't write const true = not nil; weekend puzzle or bug.

[ASerge]

WinAPI define a BOOL as an int, in this case, for example, 256 is True, but it does not fit into Boolean

[Handoko]

I do not have language knowledge outside Pascal and Basic and I rarely call WinAPI directly.

Thank you for the information.

[Thaddy]

An int holds even 32768 as true... not 256.. even -32767 is still true... Nonsense. Even MS compilers do just !=0 for true. The size of a boolean depends on the underlying library. What to expect is defined in the API documentation.

[Thaddy]

WinAPI define a BOOL as an int, in this case, for example, 256 is True, but it does not fit into Boolean

I do not have language knowledge outside Pascal and Basic and I rarely call WinAPI directly.

Thank you for the information.

That should be thank you for "alternative facts"

Note that in pascal it is dependent on the native integer type if a bool is a byte or not.
On most platforms it is a native integer type in pascal. Because the compiler processes that more efficiently.

[z505]

What is the most compatible (best) boolean to use on unix/macOS/Windows if making a library that many programs and languages will use?

The fact that a boolean can be confusing is cause for concern ;-)

I'm thinking longbool, but not sure.. just a gut hunch which could be wrong.

[Thaddy]

What is the most compatible (best) boolean to use on unix/macOS/Windows if making a library that many programs and languages will use?

The fact that a boolean can be confusing is cause for concern ;-)

I'm thinking longbool, but not sure.. just a gut hunch which could be wrong.

False is easy: that's almost always compatible and resolves to 0.
However, there is no way to guarantee what value represents true. true is not cross-platform on any system or language.
So basically you need to read documentation and use what the system asks for.
Even true := not false isn't true in all cases. True is simply not portable between languages. And that's the truth.

Then you have the choice: zero vs One or Zero vs Minus One vs True := not zero : pick one and document it.

[z505]

so sadly then, maybe just use an integer :-( and then document it

I would prefer to use something more standard like a LongBool that is available on all platforms...

I know at least what could be done for Windows systems: look at what the windows API uses and then use it.

But as for making something standard across unix/windows/macOS?  The only thing I can think of for tips, is maybe to look into OpenGL and see what they use for a boolean? As opengl runs on many systems, not just windows...

[Thaddy]

There is no standard across platforms and languages. There are standards on a per language basis, though.
FPC knows about all of them. And in FPC true and false can be re-defined because they are not keywords....

Code: Pascal  [Select][+][-]

1. const true = 10;false = 'wrong';
2. begin
3.   writeln(false);
4.   readln;
5. end.