Negativ number as true in C++.

broadpeak

Any non-zero number (positiv AND negativ number) is true in the test-condition of the "if" statement.
Where can I find the declaration of this rule in the c++ standard?

Zlatomir

Search for "Boolean conversions" in the standard document, how did you expected to work?

giesbert

Hi broadpeek,

it is very easy: an if clause checks a bool value, and bool means: false = 0.
But I can't tell you whether it is written somewhere in the standard. If it is, it's the C-standard.

broadpeak

Till now, I have known that any positiv integer is true.
And any non positiv integer and the zero are false.

Yeasterday I programmed a condition and I was surprised that negative integer is true too...

Therefore, I tried to search this definition in the C++ standard but I have found nothing about this...

Zlatomir

Quote from C++ 2003 revision:

bq. 4.12 Boolean conversions [conv.bool]
... A zero value, null pointer value, or null member pointer value is converted to false; any
other value is converted to true.

broadpeak

[quote author="Zlatomir" date="1360325618"]Quote from C++ 2003 revision:

bq. 4.12 Boolean conversions [conv.bool]
... A zero value, null pointer value, or null member pointer value is converted to false; any
other value is converted to true.[/quote]

Yeeeeeeessssss!
This was what I have searched :)

Thank you very much!

Serenity

[quote author="broadpeak" date="1360324319"]Till now, I have known that any positiv integer is true.
And any non positiv integer and the zero are false.
[/quote]

Think more in "binary" oder "hexadeciamal" mode ;)
There are no difference between positiv or negativ, because you cannot set a minus before a hexadecimal value :)

for an if clause, the "if" only see the binary or hexadecimal value.
Like
[code] if(0x80) { ...
[/code]

That could be a signed (-1 I think...) or unsigned char (127), but for the if clause, it is true :)

broadpeak

[quote author="Serenity" date="1360326276"]
[quote author="broadpeak" date="1360324319"]Till now, I have known that any positiv integer is true.
And any non positiv integer and the zero are false.
[/quote]

Think more in "binary" oder "hexadeciamal" mode ;)
There are no difference between positiv or negativ, because you cannot set a minus before a hexadecimal value :)

for an if clause, the "if" only see the binary or hexadecimal value.
Like
[code] if(0x80) { ...
[/code]

That could be a signed (-1 I think...) or unsigned char (127), but for the if clause, it is true :)

[/quote]

Ok, I understand. Thanks!

DFileccia

The explanations of this are very good. As far back as I can remember expressions evaluate as "true" when they are non-zero. In the *nix world, most functions returned 0 for success and negative for the error condition. M$ complicates this even more by making tokens to denote the system, error code, etc. Hence the need for the macro SUCCESS in a lot of their return values.

It is always a good practice, for readability and no ambiguity, to just put:

if (value == 0) or
if (value != 0)

and just let the compiler optimize it out.

Chris Kawa

[quote author="Serenity" date="1360326276"]
Think more in "binary" oder "hexadeciamal" mode ;)
There are no difference between positiv or negativ, because you cannot set a minus before a hexadecimal value :)[/quote]
Well... you can actually :) Hex is just a notation, it doesn't change the underlying bit meaning which can signify a negative integer. It's ok to do @int a = -0x1;@ , it's just not the typical use of it.

This is a historic optimization. CPUs have a special instruction for checking 0 values (like jnz in assembly), so it was (no longer true of course for quite some time) cheaper to treat "anything but zero" as true.