Hello Jonathan,
I don't see a need to do away with the restriction of no spaces in
flag_meanings, but I agree that "-", "." etc should be allowed -- I can
imagine that brackets would also be useful in some cases. In the
biodiversity data I'm dealing with I did, however, edit out the accented
letters before trying to convert it to netcdf. I suspect that for many
programmers "machine readable" does not readily extend to the full range
of accents. So I would suggest having an allowed list of characters,
e.g. a-zA-Z0-9()[]-,./_;:, or at least finding a form of words which
excludes accents,
Cheers,
Martin
> -----Original Message-----
> From: Jonathan Gregory [mailto:jonathan at met.reading.ac.uk] On Behalf
Of
> Jonathan Gregory
> Sent: 28 October 2009 10:22
> To: Juckes, Martin (STFC,RAL,SSTD)
> Cc: cf-metadata at cgd.ucar.edu; r.s.hatcher at reading.ac.uk
> Subject: [CF-metadata] Dealing with large numbers of flag values in
> netcdf cf -- what are "words"
>
> Dear Martin
>
> I would think that since words are blank-separated, they ought to be
> allowed
> to include any character except white space. The convention says "If
> multi-word
> phrases are used to describe the flag values, then the words within a
> phrase
> should be connected with underscores." To allow phrases including
> spaces, we'd
> need a different convention, such allowing flag meanings to be an
array
> of
> strings i.e. a 2D character array. I don't think that's possible in
> netCDF-
> classic for an attribute, so the attribute would have to point to a
> variable
> instead. If it is possible in netCDF-4, we could make it a feature of
> CF specific to netCDF-4. Up to now we have not used new features of
> netCDF-4 in
> the CF convention. Either way, a change to the CF standard would be
> required.
>
> Best wishes
>
> Jonathan
--
Scanned by iCritical.
Received on Wed Oct 28 2009 - 08:00:11 GMT
