Search:

I recommend using Code::Blocks with gcc instead of Borland. Code::Blocks has a wx project built in.
Download the development version, not the release.

I considered it but I was going to wrap it anyway. If I'm wrapping, I might as well wrap lower level.

I myself use Code::Blocks and gcc.
Currently writing an easy to use, platform independent C++ wrapper around Gtk+.

They are both Microsoft ONLY.

If you want to program for Linux or Mac in the future - do not use them.

Stick with OpenGL if there is a chance of someday porting your app to other operating systems.

I'm not aware of any that cost money.

DirectX==Microsoft and is NOT open source.
I don't think OpenGL is open source, it's an open standard to graphics card vendors.

My choice is OpenGL for 2 reasons.

1. I like the api.
2. This...

you're returning a local function variable...

I'm going to guess and say an environment variable must be set or something. Hopefully someone will come along who has done it.

That's a good question - one that I'd be curious to know as well.

I just started using Gtk+. It's cross platform and really, really convenient to use for automatic layout of controls. Sort of like designing a web page...

I'm writing a wrapper around it in c++...

"the whole reason i want to obtain a FILE pointer for a region of memory is so i don't have to write to a file/read in the first place"

That's what I thought. I feel your pain. One thing you can...

Actually that may not be exactly what you're looking for...

If I used MFC I'd probably consider shooting myself.

Linux:
mmap
http://www.ecst.csuchico.edu/~beej/guide/ipc/mmap.html

I had been looking for the same thing. It's easy in Linux.

I used a work-around for Windows as a temporary, really bad fix - I save to file and then read from it.

That's actually a good idea.

Not sure but I think the first fits in a float:

const float PI = 3.14159265358979323846;
const double PI_Large = 3.1415926535897932384626433832795;

#define is always const.

"With compiler optimisations and the function inlined I doubt it will make a difference"
That's true with the exception of this one:
4 * atan(1);
I doubt the compiler...

In either case, why waste processing time?

#define PI 3.14159265358979323846
#define PI_Large 3.1415926535897932384626433832795

I wouldn't depend on that. If you really need to fill the end with zero's, do it manually.

You aren't calling connect

This supports 64 and 32 bit...

inline bool IsDigit(char c)
{
return (c >= 48 && c <= 57 ? 1 : 0);
}

template < typename T >
bool ToIntegral(cchar* c, T* pIntegralType)
{

//Sec_milli is 1000

#if useLinux

timet aTime();
inline timet aTimeSec()
{
return time(0);
}

I use Code::Blocks for cross-platform development.

Yes, in that case, signed/unsigned does not matter.

I assume you want the results from just one test. Copied from above; it is the end of the char* tests and the beginning of the raw classes test.

aStrRealloc
CountTot: 68
CountEqu: 50
...