Is undefining "max" the only way to call numeric_limits<double>::max() in VS?

Hi,

For this fragment:

Code:

---

std::numeric_limits<double>::max();

---

MSVC gave me the error:

Quote:

---

1>.\dbqWrapper.cpp(109) : warning C4003: not enough actual parameters for macro 'max'
1>.\dbqWrapper.cpp(109) : error C2589: '(' : illegal token on right side of '::'
1>.\dbqWrapper.cpp(109) : error C2059: syntax error : '::'

---

It took me one hour and probably one pund of my body fat to finally try

Code:

---

#undef max
std::numeric_limits<double>::max();

---

with success.
I was not aware of the fact that the preprocessor is cabable to overwrite function names from inside the STL.
Is the #undef max the standard way of dealing with that or how do you do it?
Thank you!

That should not be necessary. Try:

Code:

---

#include <limits>
#include <iostream>

int main()
{
    std::cout << std::numeric_limits<double>::max() << std::endl;
}

---

You are right, I tried it 2 seconds ago. Seems I've included something which pulls the max macro.

You may want to #define NOMINMAX or NO_MIN_MAX (I always forget which) anyway, since som e innocuous change to the code could bring the error back. I believe NOMINMAX is Microsoft's suggested solution for this bug. You can confirm this by searching for that text on MSDN.

http://support.microsoft.com/kb/143208

Thanks a lot!

The culprit is <windows.h>, by the way. I have no idea what possessed the Win32 engineers to define these macros back then, but now backward compatibility prevents them from removing them.

Quote:

---

Originally Posted by CornedBee

The culprit is <windows.h>, by the way. I have no idea what possessed the Win32 engineers to define these macros back then, but now backward compatibility prevents them from removing them.

---

Why couldn't they just change it so that you need to define a pre-processor symbol if you want those macros, otherwise they aren't included?

Quote:

---

Originally Posted by cpjust

Why couldn't they just change it so that you need to define a pre-processor symbol if you want those macros, otherwise they aren't included?

---

You mean the other way around, surely. That way old code would break.

Quote:

---

Originally Posted by Mario F.

You mean the other way around, surely. That way old code would break.

---

Only if the old code actually used those macros. It sure wouldn't be the first time MS deprecated old "features" in their compiler. Besides, you need to draw the line at some point and say enough is enough, get rid of the useless crap.

Quote:

---

Originally Posted by cpjust

Besides, you need to draw the line at some point and say enough is enough, get rid of the useless crap.

---

Whatever you do it will damage something or force new code to new unexpected rules. It's not such a big problem really that deserves a treatment. If you look at it on the bright side, it is left there as a reminder to us all that not following the guidelines, as we often preach but don't actually do, can have that kind of outcome.

Quote:

---

Originally Posted by pheres

It took me one hour and probably one pund of my body fat to finally try

---

The error message had the word "macro" right in it...

From the MSDN link:

Quote:

---

STATUS

This behavior is by design.

---

I think they meant:

Quote:

---

STATUS

This behavior is due to bad design.

---