[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[UDUNITS #NMJ-986812]: Enhancing Udunits with ncview code?



Mary,

> We have been using udunits' utCalendar function in NCL code for some
> time now, and it's a popular function.
> 
> We get frequent requests for calendars other than the ones supported
> by udunits, like "noleap",
> "360 day", and "365 day".

The ability to offset the origin of a unit was added to
the UDUNITS package in order to accomodate units such as
Fahrenheit and Celsius.  This was done in a generic way,
and so automatically included temporal units.

Allowing that capability for non-temperature units was a
big mistake.  People are using that capability to kluge
support for physical quantities in a way that lies completely
outside the domain of units for physical quantities.  A much
better solution would be to create a library that supported
co-ordinate transformations of physical quantities directly,
rather than through kluges on their units.  Adding support for
non-Earth calendars would be adding a kluge on top of bad
existing kluge.

I understand your concern, but this is *not* the way to go.
A far better solution would be to use a calendar-conversion
library, a netCDF attribute that specified the calendar
system, and not use the "since" capability of the UDUNITS
package.

> David Pierce, the developer of ncview has created his own version of
> utCalendar that adds the above calendars to the ones that Udunits
> already supports.
> 
> Has there ever been a consideration for incorporating David's code
> into Udunits?

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: NMJ-986812
Department: Support UDUNITS
Priority: High
Status: Closed