HDF tag query

Ray Osborn ROsborn at anl.gov
Wed Sep 1 15:42:48 BST 1999


I will try and check things in F90 as soon as I get back to Argonne 
tomorrow.  The only thing that I think we should sort out before going to
1.1.1 is the treatment of dimensions in the F77 API.  I believe that
NXgetslab and NXputslab need to have the dimensions reversed (and one
subtracted from the starting dimensions) either in the Fortran or by putting
in an extra C wrapper.  Do you agree, Freddie?

I've tested that the F90 version of NXputslab, which does the dimension
conversion, s the predicted in the Fortner Browser.  I've not checked
NXgetslab on a file produced by the C API.

Regards,
Ray
--
Ray Osborn <ROsborn at anl.gov>
Materials Science Division
Argonne National Laboratory


> Mark, thanks for testing that ... i didn't have a copy of the "Fortner
> browser" around, so i hadn't got round to doing it myself. Could you also
> please check that "DFTAG_NDG" is OK as, according to the HDF docs and
> example programs, this is what we should now be using (incidentally,
> anybody know why this tag changed?)
>
> The proposal is then:
>
> * Change all DFTAG_SDS and DFTAG_SDG references to DFTAG_NDG in functions
>   that write to a nexus file
> * Leave all DFTAG_SDS and DFTAG_SDG references in functions that read
>   from a NeXus file (for backward compatability)
> * Update NeXus_version to "1.1.1" (and release)
>
> Are there any further changes that anybody would like to add before
> the next release?
>
> Regards,
>
> Freddie
>
> On Wed, 1 Sep 1999, Mark Koennecke wrote:
>
>>
>>   Freddie,
>>
>>   you were right. I tested it and changed NXgetdataID to set the tag to
>>   DFTAG_SDG. Then the linked SDS's showed up nicely in the Fortner browser
>>   which they did not do before. It never occurred to me that DFTAG_SDS
>>   means scientfic data scale and scientific data set. Abbreviations!
>>
>>   In short I suggest to change that in the general distribution.
>>
>>                   Mark Koennecke
>>
>>




More information about the NeXus-developers mailing list