Re: Cmath and math.h Changes in Xcode 2.4 (Mike Jackson)
Re: Cmath and math.h Changes in Xcode 2.4 (Mike Jackson)
- Subject: Re: Cmath and math.h Changes in Xcode 2.4 (Mike Jackson)
- From: Bill Northcott <email@hidden>
- Date: Wed, 30 Aug 2006 09:56:06 +1000
On 30/08/2006, at 8:12 AM, Mike Jackson wrote:
Looking around it seems the "safer" way to do this is to use the
#define isnan(x) std::isnan(x)
.... stuff removed
As I am learning quickly, there is no "standard" in C++ so everyone
does it
differently. So OpenSceneGraph will have to do it one way and ITK
will have
to do it another way..
There are perfectly good standards for C++. There are also a lot of
people who don't read them.
isnan() is not in those standards and it should not be in std::,
which is why a properly constructed cmath header strips it out of
math.h. Linux does not, which is why you find these problems in
Linux originated code.
The 'right way' is not to use std::isnan() in C++ code. You should
certainly not muck about with the headers because that only works on
a system like Linux where isnan() is a function. On Darwin/MacOS X
and many other systems, isnan() is a macro which translates to
different functions depending on the type of the argument.
The work around if you want isnan() functionality in C++ code is to
write a little C function say myisnan() which calls isnan(), then use
myisnan() in your C++ code. Since myisnan() will be a function not a
macro it will all work for you.
For fuller discussions of this issue see the R-devel list. For other
examples look at the sources of any C++ opensource mathematical/
statistical class library.
As you will see on the R-devel list, it took some time for the R
developers to accept that this was right, but they did in the end.
Best of luck
Bill Northcott
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden