Re: my understanding of HDF5 changes, in (guessed) completion order...

Hi Ed,

> Hello HDF Cowboys!
> 
> Here's my mental list of the changes to come in HDF5:
> 
> 1 - Dimension scales (i.e. shared dimensions)
> 
> 2 - Access by creation order index (i.e. ability to access datasets,
> dimension scales, and attributes by indexes 0, 1, 2, ..., where the
> index is also the creation order for each class of object.)
> 
> 3 - Conversion between float/int, with netcdf-like range checking.
> 
> 4 - Zero-length attributes.
> 
> 
> Anything I missed?
    Hmm, I would break change #3 into two separate changes:
        3 - Conversion between float/int

        5 - Change overflow handling in type conversions to allow fill values
            to be substituted.

    These five things are all that are currently on my list of changes
necessary, but I'm certain that others will come up... :-)

> Are all these going to be packaged into a version number? Certainly we
> will need a HDF5 version number which will be the earliest released
> version of HDF5 which works with netcdf-4. Any idea what that version
> number is going to be? 1.6.2?
    Since this is all new functionality, it's being added to the "main trunk"
of our CVS repository, currently called 1.7.x, and is slated for release in
the 1.8.0 release of HDF5.  At some point, probably soon (when Ray finishes
up the int<->float conversion work), you'll want to transition to using
1.7.x snapshots for your prototyping work, so we can get some feedback and
verification of the changes.

    Quincey
>From owner-netcdf-hdf@xxxxxxxxxxxxxxxx 12 2003 Nov -0700 09:56:24 
Message-ID: <wrx7k25cypz.fsf@xxxxxxxxxxxxxxxxxxxxxxx>
Date: 12 Nov 2003 09:56:24 -0700
From: Ed Hartnett <ed@xxxxxxxxxxxxxxxx>
To: netcdf-hdf@xxxxxxxxxxxxxxxx
Subject: something startling I just noticed...
Received: (from majordo@localhost)
        by unidata.ucar.edu (UCAR/Unidata) id hACGuQs1026829
        for netcdf-hdf-out; Wed, 12 Nov 2003 09:56:26 -0700 (MST)
Received: from rodney.unidata.ucar.edu (rodney.unidata.ucar.edu 
[128.117.140.88])
        by unidata.ucar.edu (UCAR/Unidata) with ESMTP id hACGuPOb026824
        for <netcdf-hdf@xxxxxxxxxxxxxxxx>; Wed, 12 Nov 2003 09:56:25 -0700 (MST)
Organization: UCAR/Unidata
Keywords: 200311121656.hACGuPOb026824
Lines: 26
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: owner-netcdf-hdf@xxxxxxxxxxxxxxxx
Precedence: bulk

Hello all!

If you HDF5 guys implement the indexing and shared dimensions, we will
be able to create and read a HDF5 file using the netcdf-4 interface
with no special attributes that need to be understood.

If that's true, can we use the netcdf-4 interface to read *other* HDF5
files, even though they weren't created with netcdf-4? The only
impediments are the indexing of datasets, attributes, and shared
dimensions in creation order.

If all HDF5 files had this indexing, We might very easily be able to
operate on *any* HDF5 file. What do you think Quincey?

And, if it comes right down to it, why not fake it?  Quincey, if I
iterate through the list of datasets, I'll get them in alphabetical
order? Is that correct?

As long as I provide a consistent numbering scheme, it seems like
most software (eg. grads) that uses netcdf might well be able to read
HDF5 files that were created before this project even started, though
they obviously wouldn't have shared or named dimensions.

Just a thought...

Ed