r.timestamp: support also YYYY-DOY

[neteler]

It would be of great help (esp. for time series) to add year/doy (Day Of Year) support to r.timestamp. From this t.register would benefit as well.

The "only" change needed seems to be an upgrade of G_scan_timestamp():

​http://grass.osgeo.org/programming7/scan_8c_source.html#l00043

probably by adding a new function scan_doy() Day Of Year

[wenzeslaus]

If doing this, try to follow some consistent terminology/naming.

I've recently encountered different namings in GRASS: DOY, Day of Year, day of year, doy, daynum and day.

[huhabla]

This must be implemented in the C datetime library and in the temporal framework parser in datetime_math.py[1]. I hoped that the dateutils[2] already support this to have it out of the box, but unfortunately they don't.

Hence we should orient on the ISO 8601 standard for ordinal dates when implementing this (YYYY-DDD) but not YYYYDDD, since the temporal framework and the C datetime library also support relative time and can't decide if 1900031 is 31. Jan. 1900 or if it is the number of seconds, days, ... .

[1] ​http://grass.osgeo.org/programming7/datetime__math_8py_source.html#l00704 [2] ​https://labix.org/python-dateutil

[neteler]

See also #964

[martinl]

Milestone renamed

[neteler]

Ticket retargeted after milestone closed

[martinl]

Ticket retargeted after milestone closed

[martinl]

Ticket retargeted after milestone closed