Use the same buffer for separate DATE and TIME - mysql

Question

I am using the MySQL C api along with prepared statements. When issuing a INSERT query with a TIME field the result is in HHH:MM:SS format, where the hours are the amount of hours elapsed this month. For instance if the date is 2015-02-21, and the time is 21:30:00 the time would be displayed as 525:30:00 but I want to use the HH:MM:SS format instead (e.g 21:30:00), which would be the actual time of the day.
sbind[3].buffer_type=MYSQL_TYPE_DATE;
sbind[3].buffer= (char *)&ts; // Pointer to a MYSQL_TIME data structure
sbind[3].is_null= 0;
sbind[3].length= 0;
sbind[4] = sbind[3];
sbind[4].buffer_type=MYSQL_TYPE_TIME;
mysql_stmt_bind_param(stmt, sbind); // sbind is an array of MYSQL_BIND structures
ts.year= 1900+tm_info->tm_year; // tm_info is a pointer to a tm structure
ts.month= 1+tm_info->tm_mon;
ts.day= tm_info->tm_mday;
ts.hour= tm_info->tm_hour;
ts.minute= tm_info->tm_min;
ts.second= tm_info->tm_sec;
This code will prepare the date field as yyyy-mm-dd and fill it with the date in tm_info. Likewise it will do the same thing for the time field but in the HHH:MM:SS format.
A unfashionable way which works is to use a separate MYSQL_TIME structure for the time, but I aim for a more elegant way to handle this.
(EDIT: Here I have included the relevant client side code
MYSQL_TIME ts;
MYSQL_STMT *stmt;
MYSQL_BIND sbind[2];
...
char query[QUERY_BUFFER_SIZE];
strcpy(query, "INSERT INTO `mytable` (date,time) VALUES(?,?)");
if(mysql_stmt_prepare(stmt, query, strlen(query))){
return mysql_stmt_errno(stmt);
}
...
time_t rawtime;
time(&rawtime); // get current time
struct tm *tm_info = localtime ( &rawtime );
...
memset(sbind,0,sizeof(sbind));
sbind[0].buffer_type=MYSQL_TYPE_DATE;
sbind[0].buffer= (char *)&ts; // Pointer to a MYSQL_TIME data structure
sbind[0].is_null= 0;
sbind[0].length= 0;
sbind[1] = sbind[0];
sbind[1].buffer_type=MYSQL_TYPE_TIME;
mysql_stmt_bind_param(stmt, sbind); // sbind is an array of MYSQL_BIND structures
ts.year= 1900+tm_info->tm_year; // tm_info is a pointer to a tm structure
ts.month= 1+tm_info->tm_mon;
ts.day= tm_info->tm_mday;
ts.hour= tm_info->tm_hour;
ts.minute= tm_info->tm_min;
ts.second= tm_info->tm_sec;
if(mysql_stmt_execute(stmt)){
return mysql_stmt_errno(stmt);
}
This assumes mysql is a valid connection. The table mytable in this case only contains a DATE type and a TIME type.
)

Answer 1

That's the natural representation for the TIME type - it, as the name suggests, only holds time, and this is the natural way to express time larger than a day. It does, as the documentation suggests, use HH:MM:SS for smaller values.
Since TIME does not care about shenanigans like leap seconds, to exclude full days, just take tm_hour%24. But, to allow for shenanigans like DST transitions, you have nothing to do but add the TIME to the starting point of the specific month and do DATETIME arithmetic with the stock functions.
#c45602234:
Trying to outsmart libmysql FAILED (mysql-connector-c-6.1.5/libmysql/libmysql.c:1964):
static void store_param_date(NET *net, MYSQL_BIND *param)
{
MYSQL_TIME tm= *((MYSQL_TIME *) param->buffer);
tm.hour= tm.minute= tm.second= tm.second_part= 0;
net_store_datetime(net, &tm);
}
As you can see, it always uses up the entire structure (apparently, so that relevant functions always work as expected).