Dear all,
as I am the "at least one other person" to whom Nan refered, let me clarify my position:
1) I would strongly argue against adding another way of formatting time through the backdoor via a standard_name.
2) I do see quite a bit of sense in re-modeling the date and time handling in netcdf-CF, because the current exclusive "something since ..." does have its drawbacks and makes things sometimes more complicated than they would need to be. I would just like to point out the discussions about climatological time axes, paleo dates, etc. Furthermore, interoperable services typically rely on ISO metadata and handle ISO 8601 time stamps. Thus, in order to translate back and forth to and from netcdf-CF one always needs to convert. Yes! There are libraries that can do this, but in the life of a scientist, these libraries are not always readily at hand, or it takes time to find out how they work, and your favorite programmer is on vacation.
3) Alexandar has rightly pointed out that there seems no easy consensus on this list to add ISO 8601 to the convention. Well, one reason for this is that a convention change must be discussed in a track ticket (see
http://cf-pcmdi.llnl.gov/discussion/about-the-cf-trac-ticket-system). The other reason (more important) is (as has been pointed out by others) that this is actually an issue that goes beyond the CF realm and affects the netcdf library itself. I found it very interesting to see that the Unidata CDM has been expanded to allow for ISO-like datetime values. Indeed the question is: when will this be fully supported by the netcdf library and the ncdump/ncgen tools? As soon as this is the case, I would in fact advocate strongly that we open a track ticket and let the convention follow what is done in netcdf and the CDM.
There is some interesting reading at
http://www.dwd.de/bvbw/generator/DWDWWW/Content/Oeffentlichkeit/TI/TI1/Informationstechnik/Software/UniDart/downloads/2__XML__Datatypes__download,templateId=raw,property=publicationFile.pdf/2_XML_Datatypes_download.pdf -- specifically this document declares datatypes such as gYear, gYearMonth, gMonthDay for recurring dates (remember this climatology issue and the ban on using "month" as period interval?).
So, I clearly see a future here - but it's first in the hands of Unidata, and then probably more for CF 1.7.
Best regards,
Martin
------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
Forschungszentrum Juelich GmbH
52425 Juelich
Sitz der Gesellschaft: Juelich
Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender),
Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
Prof. Dr. Sebastian M. Schmidt
------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
Received on Thu Mar 21 2013 - 07:57:59 GMT
