NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.

To learn about what's going on, see About the Archive Site.

Re: questions about attributes and compound types...

NOTE: The netcdf-hdf mailing list is no longer active. The list archives are made available for historical reasons.

Hi John,

> >>>>>>We've got a question: is there a way to attach an attribute to a
> >>>>>>member of a compound type dataset. That is, if I have a dataset of
> >>>>>>something that is a compound type, can I attach at attribute not to
> >>>>>>the whole dataset, but just to one member of the compound type?
> >>>>>>       
> >>>>>>
> >>>>>>            
> >>>>>>
> >>>>>   Hmm, we've had a couple of requests for similar features, but you 
> >>>>> can't
> >>>>>do this currently.  What's the motivation for this?
> >>>>>
> >>>>>     
> >>>>>
> >>>>>          
> >>>>>
> >>>>I think I will let John answer this part...
> >>>>   
> >>>>
> >>>>        
> >>>>
> >>the motivation comes from creating a unifying model with the opendap 
> >>data model.
> >>
> >>opendap structures contain "first class" variables that can have attributes.
> >>    
> >>
> >    Hmm, interesting...
> >
> >  
> >
> >>however, if thats not feasible, i can do a workaround where opendap 
> >>structures become groups instead.
> >>    
> >>
> >    Well, it definitely isn't currently supported.  Should I open an issue 
> > for
> >this in our tracker so we consider in the future?  I'm just a bit concerned
> >about supporting the full generality of it, like attributes on parts of array
> >datatypes or nested compound datatypes, etc.
> >
> >    Quincey
> >  
> >
> yes, i think its worth considering in the future. this would make 
> structure members first class citizens. the attributes would apply to 
> the structure member as a whole, not individual array elements, just as 
> for an array dataset.
    Ok, I've entered an issue in our bugzilla.

> as i said, the motivation comes from opendap structures. ironically, i 
> suspect that it was at least in part motivated there by mapping hdf 
> (4/5?) into opendap, where they decided to map "groups" into 
> "structures". so we can just undo that and map opendap structures to 
> groups when needed.
    That would probably work OK.

    Quincey

  • 2004 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the netcdf-hdf archives: