Re: [Boost-users] [date-time] boost::time_duration wrapping

On Aug 6, 2014, at 7:07 PM, Justin Muncaster wrote:

Hello,

I’m having issues with wrapping of boost::time_duration derived types. I can’t find anywhere in the documentation for when the time_duration types are supposed to wrap around.

If the duration types are based on signed integers (which I believe they are), then "wrapping around” is undefined behavior. Once you’ve invoked undefined behavior, then you can’t expect *anything*, or, as I say, “there are no wrong answers” — Marshall

I have attached some sample source code below. The output is: SecCount = 80000000 is OK. Strange Stranger Problematic SecCount = 80000001 is STRANGE.

I expect the output to be SecCount = 80000000 is OK. SecCount = 80000001 is OK.

The problem we ran into had to do with seconds and milliseconds having different maximum permissible values. Any idea why boost::posix_time::seconds goes negative at 0x80000001 while milliseconds is not until much later?

Thanks,

Justin

// // test.cpp //

#include #include <iostream>

using namespace std; using namespace boost::posix_time;

//------------------------------------------------------- //------------------------------------------------------- void Strange(long SecCount) { long MilliCount = 1000 * SecCount;

bool IsStrange = false;

if (SecCount >= 0 && seconds(SecCount).is_negative()) { cout << "Strange" << endl; IsStrange = true;

if (!milliseconds(MilliCount).is_negative()) { // If seconds wraps, why doesn't milliseconds? cout << "Stranger" << endl; IsStrange = true; } }

if (seconds(SecCount) != milliseconds(MilliCount)) { cout << "Problematic" << endl; IsStrange = true; }

if (!IsStrange) { cout << "SecCount = " << SecCount << " is OK." << endl; } else { cout << "SecCount = " << SecCount << " is STRANGE." << endl; } }

//------------------------------------------------------- //------------------------------------------------------- int main() { cout << hex; Strange(0x80000000); Strange(0x80000001); return 0; }

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users