Reputation: 77
I have this piece of code in my application:
void dateToString(char * epoch, struct tm * data) {
int year, month, day, hour, minute, second;
year = data->tm_year + 1900;
month = data->tm_mon + 1;
day = data->tm_mday;
hour = data->tm_hour;
minute = data->tm_min;
second = data->tm_sec;
sprintf(epoch, "%04d-%02d-%02dT%02d:%02d:%02d.0Z", year, month, day, hour, minute, second);
}
...
(in main function)
...
currentDate = gmtime(¤tTime);
dateToString(epoch, currentDate);
sprintf(timeIntMs, "%05u", tiMs);
measFile << epoch << "," << measBuffer; measFile.flush();
where:
ofstream measFile;
unsigned int tiMs;
struct tm * currentDate; // to storage current time [dd-mm-yyyy h:m:s]
char measBuffer[256];
time_t currentTime;
char epoch[23];
char timeIntMs[5];
The problem is that in this case in my measFile
output file the value of epoch is empty (i.e. string with length 0), while if I change the order of the calls like this:
currentDate = gmtime(¤tTime);
dateToString(epoch, currentDate);
measFile << epoch << "," << measBuffer; measFile.flush();
sprintf(timeIntMs, "%05u", tiMs);
i.e. by moving the sprintf
call after the saving of the file the variable epoch maintains its own value (e.g. 2015-11-08T11:06:05.0Z
).
What's wrong in this code?
Upvotes: 0
Views: 331
Reputation: 141554
You are writing out of bounds with timeIntMs
.
Instead do char timeIntMs[6];
and then sprintf(timeIntMs, "%05u", tiMs % 100000);
. Read about null termination; and the 5
in %05u
is a minimum field width, not a maximum.
Preferably use snprintf
instead, as then even if you make a mistake in arithmetic, you won't cause a buffer overflow.
Upvotes: 4