[boost][chrono] time_since_epoch different on windows and linux

[Ralf Globisch]

Hi,

I've run into a bit of a funny one and I'm not sure if I'm missing
something or whether I've encountered a bug.

The following code excerpt *seems* to have 2 different results on
windows 7 (VS2010) and linux (gcc4.6) using boost version 1.47.

#include <iostream>
#include <cstdint>
#include <boost/chrono.hpp>

int main(int argc, char** argv)
{
boost::chrono::duration<double> sec =
boost::chrono::system_clock::now().time_since_epoch();
uint32_t uiSeconds = sec.count();
double dMinutes = uiSeconds/60.0;
double dHours = dMinutes/60.0;
double dDays = dHours/24.0;
double dYears = dDays/365.0;

std::cout << "Sec: " << uiSeconds <<
" Min: " << dMinutes <<
" Hours: " << dHours <<
" Days: " << dDays <<
" Years: " << dYears << std::endl;

return 0;
}


This outputs the following:
Linux:
Sec: 1323362917 Min: 2.2056e+07 Hours: 367601 Days: 15316.7 Years: 41.9636

Windows:
Sec: 82934819 Min: 1.38225e+006 Hours: 23037.4 Days: 959.894 Years: 2.62985

The linux results is the expected result: 1970 + 41 years = ~2011.
Why does the windows result differ? I've run the windows version
on two different windows machines with the same result.

Am I using the library incorrectly? I'm trying to obtain seconds since 1970
so that I can from there calculate seconds since 1900 to calculate
an NTP timestamp.

Thanks for any help/suggestions!
Ralf

[Klaim - Joël Lamotte]

Hi,

The documentation makes clear that epoch is implementation dependent :

"The library leaves epochs unspecified."

From : http://www.boost.org/doc/libs/1_48_0/doc/html/chrono.html

"In each of the examples above, a different epoch is implied. Sometimes
an epoch has meaning for several millennia. Other times the meaning of an
epoch is lost after a while (such as the start of a timer, or when the
computer booted). However, if two
time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#chrono.reference.cpp0x.time_point_hpp.time_point>s
are known to share the same epoch, they can be subtracted, yielding a valid
duration<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#chrono.reference.cpp0x.duration_hpp.duration>,
even if the definition of the epoch no longer has meaning.

In *Boost.Chrono*, an epoch is a purely abstract and unspecified concept.
There is no type representing an epoch. It is simply an idea that relates
(or doesn't) time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#chrono.reference.cpp0x.time_point_hpp.time_point>s
to a clock, and in the case that they share a clock,
time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#chrono.reference.cpp0x.time_point_hpp.time_point>s
to one another.
time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#chrono.reference.cpp0x.time_point_hpp.time_point>s
associated with different clocks are generally not interoperable unless the
relationship between the epochs associated with each clock is known."

From http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/users_guide.html

My understanding is that chrono isn't meant to provide absolute time
"position", but time measurement and manipulations.
The difference being that epoch is an arbitrary referential used to allow
now() to be written and worked with.

Joël Lamotte

[Ralf Globisch]

Post by Klaim - Joël Lamotte

On 08/12/2011 at 23:57, in message

<CAOU91OOrXc_BkOcFt1fK=e+oC5VkC-***@mail.gmail.com>,
Klaim -

Post by Klaim - Joël Lamotte
Hi,
"The library leaves epochs unspecified."

From : http://www.boost.org/doc/libs/1_48_0/doc/html/chrono.html

"In each of the examples above, a different epoch is implied. Sometimes
an epoch has meaning for several millennia. Other times the meaning of an
epoch is lost after a while (such as the start of a timer, or when the
computer booted). However, if two

time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.htm

Post by Klaim - Joël Lamotte
l#chrono.reference.cpp0x.time_point_hpp.time_point>s
are known to share the same epoch, they can be subtracted, yielding a valid

duration<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.html#

Post by Klaim - Joël Lamotte
chrono.reference.cpp0x.duration_hpp.duration>,
even if the definition of the epoch no longer has meaning.
In *Boost.Chrono*, an epoch is a purely abstract and unspecified concept.
There is no type representing an epoch. It is simply an idea that relates
(or doesn't)

time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.htm

Post by Klaim - Joël Lamotte
l#chrono.reference.cpp0x.time_point_hpp.time_point>s
to a clock, and in the case that they share a clock,

time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.htm

Post by Klaim - Joël Lamotte
l#chrono.reference.cpp0x.time_point_hpp.time_point>s
to one another.

time_point<http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/reference.htm

Post by Klaim - Joël Lamotte
l#chrono.reference.cpp0x.time_point_hpp.time_point>s
associated with different clocks are generally not interoperable unless the
relationship between the epochs associated with each clock is

known."

Post by Klaim - Joël Lamotte

From

http://www.boost.org/doc/libs/1_48_0/doc/html/chrono/users_guide.html

Post by Klaim - Joël Lamotte
My understanding is that chrono isn't meant to provide absolute time
"position", but time measurement and manipulations.
The difference being that epoch is an arbitrary referential used to allow
now() to be written and worked with.
Joël Lamotte

Thanks a lot for the clarification Joël, I had read the quoted section,

but interpreted it incorrectly.

Regards,
Ralf

[Vicente J. Botet Escriba]

Post by Ralf Globisch
Hi,
I've run into a bit of a funny one and I'm not sure if I'm missing
something or whether I've encountered a bug.
The following code excerpt *seems* to have 2 different results on
windows 7 (VS2010) and linux (gcc4.6) using boost version 1.47.
#include<iostream>
#include<cstdint>
#include<boost/chrono.hpp>
int main(int argc, char** argv)
{
boost::chrono::duration<double> sec =
boost::chrono::system_clock::now().time_since_epoch();
uint32_t uiSeconds = sec.count();
double dMinutes = uiSeconds/60.0;
double dHours = dMinutes/60.0;
double dDays = dHours/24.0;
double dYears = dDays/365.0;
std::cout<< "Sec: "<< uiSeconds<<
" Min: "<< dMinutes<<
" Hours: "<< dHours<<
" Days: "<< dDays<<
" Years: "<< dYears<< std::endl;
return 0;
}
Sec: 1323362917 Min: 2.2056e+07 Hours: 367601 Days: 15316.7 Years: 41.9636
Sec: 82934819 Min: 1.38225e+006 Hours: 23037.4 Days: 959.894 Years: 2.62985
The linux results is the expected result: 1970 + 41 years = ~2011.
Why does the windows result differ? I've run the windows version
on two different windows machines with the same result.
Am I using the library incorrectly? I'm trying to obtain seconds since 1970
so that I can from there calculate seconds since 1900 to calculate
an NTP timestamp.
Thanks for any help/suggestions!

Hi,

time_since_epoch() gives the time since an undetermined epoch. I will use instead
boost::chrono:time_system_clock::to_time_t which gives you the number of nanoseconds since 1970.

std::time_t sec =
boost::chrono:time_system_clock::to_time_t(
boost::chrono::system_clock::now().time_since_epoch()
);


Hth,
Vicente

[Ralf Globisch]

"Vicente J.

Post by Ralf Globisch

Post by Ralf Globisch
Hi,
I've run into a bit of a funny one and I'm not sure if I'm missing
something or whether I've encountered a bug.
The following code excerpt *seems* to have 2 different results on
windows 7 (VS2010) and linux (gcc4.6) using boost version 1.47.
#include<iostream>
#include<cstdint>
#include<boost/chrono.hpp>
int main(int argc, char** argv)
{
boost::chrono::duration<double> sec =
boost::chrono::system_clock::now().time_since_epoch();
uint32_t uiSeconds = sec.count();
double dMinutes = uiSeconds/60.0;
double dHours = dMinutes/60.0;
double dDays = dHours/24.0;
double dYears = dDays/365.0;
std::cout<< "Sec: "<< uiSeconds<<
" Min: "<< dMinutes<<
" Hours: "<< dHours<<
" Days: "<< dDays<<
" Years: "<< dYears<< std::endl;
return 0;
}
Sec: 1323362917 Min: 2.2056e+07 Hours: 367601 Days: 15316.7 Years: 41.9636
Sec: 82934819 Min: 1.38225e+006 Hours: 23037.4 Days: 959.894

2.62985

Post by Ralf Globisch
The linux results is the expected result: 1970 + 41 years = ~2011.
Why does the windows result differ? I've run the windows version
on two different windows machines with the same result.
Am I using the library incorrectly? I'm trying to obtain seconds since 1970
so that I can from there calculate seconds since 1900 to calculate
an NTP timestamp.
Thanks for any help/suggestions!

Hi,
time_since_epoch() gives the time since an undetermined epoch. I will

use

Post by Ralf Globisch
instead
boost::chrono:time_system_clock::to_time_t which gives you the number

of

Post by Ralf Globisch
nanoseconds since 1970.
std::time_t sec =
boost::chrono:time_system_clock::to_time_t(
boost::chrono::system_clock::now().time_since_epoch()
);
Hth,
Vicente
_______________________________________________
Boost-users mailing list
http://lists.boost.org/mailman/listinfo.cgi/boost-users

Hi Vicente,

That helps a lot, thanks!

Regards,
Ralf

[Ralf Globisch]

"Vicente J.

Post by Ralf Globisch

Post by Ralf Globisch
Hi,
I've run into a bit of a funny one and I'm not sure if I'm missing
something or whether I've encountered a bug.
The following code excerpt *seems* to have 2 different results on
windows 7 (VS2010) and linux (gcc4.6) using boost version 1.47.
#include<iostream>
#include<cstdint>
#include<boost/chrono.hpp>
int main(int argc, char** argv)
{
boost::chrono::duration<double> sec =
boost::chrono::system_clock::now().time_since_epoch();
uint32_t uiSeconds = sec.count();
double dMinutes = uiSeconds/60.0;
double dHours = dMinutes/60.0;
double dDays = dHours/24.0;
double dYears = dDays/365.0;
std::cout<< "Sec: "<< uiSeconds<<
" Min: "<< dMinutes<<
" Hours: "<< dHours<<
" Days: "<< dDays<<
" Years: "<< dYears<< std::endl;
return 0;
}
Sec: 1323362917 Min: 2.2056e+07 Hours: 367601 Days: 15316.7 Years: 41.9636
Sec: 82934819 Min: 1.38225e+006 Hours: 23037.4 Days: 959.894

2.62985

Post by Ralf Globisch
The linux results is the expected result: 1970 + 41 years = ~2011.
Why does the windows result differ? I've run the windows version
on two different windows machines with the same result.
Am I using the library incorrectly? I'm trying to obtain seconds since 1970
so that I can from there calculate seconds since 1900 to calculate
an NTP timestamp.
Thanks for any help/suggestions!

Hi,
time_since_epoch() gives the time since an undetermined epoch. I will

use

Post by Ralf Globisch
instead
boost::chrono:time_system_clock::to_time_t which gives you the number

of

Post by Ralf Globisch
nanoseconds since 1970.
std::time_t sec =
boost::chrono:time_system_clock::to_time_t(
boost::chrono::system_clock::now().time_since_epoch()
);
Hth,
Vicente
_______________________________________________
Boost-users mailing list
http://lists.boost.org/mailman/listinfo.cgi/boost-users

Have verified that I get the correct time now.
Just a minor correction in case someone
stumbles on this thread and fails to compile:

It should be
std::time_t sec =
boost::chrono:system_clock::to_time_t(
boost::chrono::system_clock::now()
);

Thanks for writing and maintaining this wonderful library Vicente.

Regards,
Ralf