asked on Getting GMT time into a CTime
When our server passes time/date values to clients, it does so in a formatted string representing the time in GMT (UTC). It's easy to get UTC from a CTime object with FormatGmt(), but there does not seem to be a way to put UTC back into another CTime object after we parse that string. Any time that I give to CTime seems to be interpreted as local time.
A clumbsy workaround is to store the original values of _timezone and _daylight (global variables in C runtime), set them to 0, construct my CTime object, then restore the variables. This works, but is ugly and obviously not threadsafe.
Can anyone offer a better alternative?
In short, I'm looking for a function like:
CTime GetCTimeFromGmt(int year, int month, int day, int hour, int minute, int second)
{
}
where the passed parameters are relative to GMT. The function must be thread safe.
Thanks,
Matthew
A clumbsy workaround is to store the original values of _timezone and _daylight (global variables in C runtime), set them to 0, construct my CTime object, then restore the variables. This works, but is ugly and obviously not threadsafe.
Can anyone offer a better alternative?
In short, I'm looking for a function like:
CTime GetCTimeFromGmt(int year, int month, int day, int hour, int minute, int second)
{
}
where the passed parameters are relative to GMT. The function must be thread safe.
Thanks,
Matthew
System Programming
Last Comment
azami
ASKER CERTIFIED SOLUTION
THIS SOLUTION ONLY AVAILABLE TO MEMBERS.
View this solution by signing up for a free trial.
Members can start a 7-Day free trial and enjoy unlimited access to the platform.
ASKER
The ripple effect when we cross a day boundary (need to see if the day is right, and the month, and the year...) Seems a bit much. But this seems to work:
CTime GetCTimeFromGmt(int year, int month, int day, int hour, int minute, int second)
{
struct tm gmt_struct;
gmt_struct.tm_sec = second;
gmt_struct.tm_min = minute;
gmt_struct.tm_hour = hour;
ASSERT(day >= 1 && day <= 31);
gmt_struct.tm_mday = day;
ASSERT(month >= 1 && month <= 12);
gmt_struct.tm_mon = month - 1; // tm_mon is 0 based
ASSERT(year >= 1900);
gmt_struct.tm_year = year - 1900; // tm_year is 1900 based
gmt_struct.tm_isdst = -1; //DST is unknown
time_t gmt_tt = mktime(&gmt_struct);
// convert to gmt as though it was local time to get the offset
struct tm *reversed_offset_struct = gmtime( &gmt_tt );
time_t reversed_offset_tt = mktime(reversed_offset_str
time_t offset = reversed_offset_tt - gmt_tt;
time_t local_tt = gmt_tt - offset;
return CTime(local_tt);
}
My only concern now is the thread safety of gmtime(), which uses a static buffer. I think the MT CRT solves this, but I haven't been able to confirm one way or another.
I have to admit that I'm a bit surprised that neither CTime nor the C runtime offer a way to use GMT time as input!
Thanks for your help Migel. I want to give this a few days to see whether anyone can comment on the thread safety before I accept your answer.
-Matthew
CTime GetCTimeFromGmt(int year, int month, int day, int hour, int minute, int second)
{
struct tm gmt_struct;
gmt_struct.tm_sec = second;
gmt_struct.tm_min = minute;
gmt_struct.tm_hour = hour;
ASSERT(day >= 1 && day <= 31);
gmt_struct.tm_mday = day;
ASSERT(month >= 1 && month <= 12);
gmt_struct.tm_mon = month - 1; // tm_mon is 0 based
ASSERT(year >= 1900);
gmt_struct.tm_year = year - 1900; // tm_year is 1900 based
gmt_struct.tm_isdst = -1; //DST is unknown
time_t gmt_tt = mktime(&gmt_struct);
// convert to gmt as though it was local time to get the offset
struct tm *reversed_offset_struct = gmtime( &gmt_tt );
time_t reversed_offset_tt = mktime(reversed_offset_str
time_t offset = reversed_offset_tt - gmt_tt;
time_t local_tt = gmt_tt - offset;
return CTime(local_tt);
}
My only concern now is the thread safety of gmtime(), which uses a static buffer. I think the MT CRT solves this, but I haven't been able to confirm one way or another.
I have to admit that I'm a bit surprised that neither CTime nor the C runtime offer a way to use GMT time as input!
Thanks for your help Migel. I want to give this a few days to see whether anyone can comment on the thread safety before I accept your answer.
-Matthew
Hi!
IMHO all CRT routimes must be thread safe!
look at CRT source gmtime.c (...\Program Files\Microsoft Visual Studio\VC98\CRT\SRC\) and you founs that it isn`t exception from mainstream :-)
IMHO all CRT routimes must be thread safe!
look at CRT source gmtime.c (...\Program Files\Microsoft Visual Studio\VC98\CRT\SRC\) and you founs that it isn`t exception from mainstream :-)
ASKER
I finally got around to installing the CRT source code to check. I agree that the CRT _should_ be thread-safe, but unfortunately, not all of it is. Fortunately, gmtime's buffer _is_ threadsafe, so the code (with my modifications) should work until the Y2.038K bug strikes. BTW, _timezone and _daylight are among the globals that do not seem to have the thread-safety harness, so my original code that temporarily changed those (see question) was NOT threadsafe.
CTime GetCTimeFromGmt(int year, int month, int day, int hour, int minute, int second)
{
struct tm gmt_struct;
gmt_struct.tm_sec = second;
gmt_struct.tm_min = minute;
gmt_struct.tm_hour = hour;
ASSERT(day >= 1 && day <= 31);
gmt_struct.tm_mday = day;
ASSERT(month >= 1 && month <= 12);
gmt_struct.tm_mon = month - 1; // tm_mon is 0 based
ASSERT(year >= 1900);
gmt_struct.tm_year = year - 1900; // tm_year is 1900 based
gmt_struct.tm_isdst = -1; //DST is unknown
time_t gmt_tt = mktime(&gmt_struct);
// convert to gmt as though it was local time to get the offset
struct tm *reversed_offset_struct = gmtime( &gmt_tt );
time_t reversed_offset_tt = mktime(reversed_offset_str
time_t offset = reversed_offset_tt - gmt_tt;
time_t local_tt = gmt_tt - offset;
return CTime(local_tt);
}
My only concern now is the thread safety of gmtime(), which uses a static buffer. I think the MT CRT solves this, but I haven't been able to confirm one way or another.
I have to admit that I'm a bit surprised that neither CTime nor the C runtime offer a way to use GMT time as input!
Thanks for your help Migel. I want to give this a few days to see whether anyone can comment on the thread safety before I accept your answer.
-Matthew