⇐ ⇒

[CF-metadata] time as ISO strings

From: John Caron <caron>
Date: Fri, 29 Oct 2010 17:35:29 -0600

On 10/26/2010 8:41 AM, Nan Galbraith wrote:
> Hi all -
>
> We seem to be flogging this more than necessary,
> so sorry if I'm just adding to that.
>
> On 10/25/10 3:43 PM, Ben Hetland wrote:
>> I do believe, though, that ISO date/time strings do have a utility
>> in--or as part of-- textual metadata fields associated with such
>> coordinates, for instance as in the unit property of the time dimension
>> ("minutes since 2001-01-01T00Z").
> I don't agree that we want to change the convention
> for the time units ("minutes since 2001-01-01T00Z"), at
> least not until - or if - UDUNITS decides to implement
> ISO date strings. UDUNITS is too intrinsic to CF, and
> changing the convention for time has serious consequences
> for existing data sets. Time's units are not really
> 'textual metadata fields' - they're intrinsic to
> (machine) understanding of the time coordinate.
>
> I DO think ISO strings are useful in other metadata
> fields, where CF doesn't provide a standard. This means
> that provenance date strings and data set time limits in
> my global attributes are in a different format from my
> time variable's unit's reference date string, but that's
> a small price to pay, IMHO, because these are used very
> differently.
>
> Cheers - Nan
>

A bit more flogging I suppose:

1) no one is suggesting to replace the udunits convention; the original
suggestion a few years back was to allow ISO 8601 formatted time strings
in addition.

2) Udunits accepts ISO Strings for the date part, AFAIU, but allows more
variations I think.

The formal grammar is here:

http://www.unidata.ucar.edu/software/udunits/udunits-2/udunits2lib.html#Grammar

I assume this specification is what we want to keep as the "standard",
in order to keep backwards compatibility.

3) The problem with udunits time handling is that it doesnt handle
anything other than the default Calendar. So, it cant be used as the
reference implementation for non-standard Calendars.

So we should decide on the reference implementation for non-standard
Calendars. Perhaps we can just say "cflib" ?? Ed, what are using under
the covers for that?
Received on Fri Oct 29 2010 - 17:35:29 BST

This archive was generated by hypermail 2.3.0 : Tue Sep 13 2022 - 23:02:41 BST

⇐ ⇒