[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 20030423:NetCDF 3.5 to HDF5

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: 20030423:NetCDF 3.5 to HDF5
  • Date: Wed, 23 Apr 2003 13:59:36 -0600
>To: address@hidden
>From: Robert VINCENT <address@hidden>
>Subject: From NetCDF 3.5 to HDF5
>Organization: UCAR/Unidata
>Keywords: 200304231819.h3NIJM7U024784

Hi Bert,

> I'm in the process of extending some code to support HDF5 as well as
> NetCDF.  Are there any notes you've created to address generic issues that
> may arise in this conversion?

No, sorry.  We're about to embark on a two-year project providing an
extended netCDF-4 interface over the HDF5 file format, but I don't
think there's anything in the proposed work that would help you now.

> I'm especially concerned about converting code that uses "ncvarputg",
> "nc_put_varm", or other functions that can specify a NetCDF "imap".  Is
> there any equivalent in HDF5 that you know of?

No, but I don't think that should be of much concern, unless you
actually have users using ncvarputg or nc_put_varm functions.  In my
opinion, these interfaces were an example of over-generalizing the
interface.  The extra generality they support is rarely used or
needed.  When we wrote a Java interface, we didn't bother trying to
provide similar functionality, because we couldn't provide any
non-contrived examples where it would be useful.  Similarly, I didn't
include analogous "mapped" interfaces for the C++ netCDF interface,
but no one has ever missed them or asked for them.

It would be possible to implement this kind of interface on top of
HDF5 with wrappers that used loops to access the individual elements
and reorder them, but HDF5's struct accesses are a simpler and more
useful generalization, in my opinion.

If you know of any practical uses of the ncvarputg or nc_put_varm type
interfaces, I'd be interested in hearing about them.  We've answered
thousands of netCDF support questions, but probably only a handful
have dealt with any of those interfaces, and we usually recommended
simpler ways to accomplish the task.

--Russ

_____________________________________________________________________

Russ Rew                                         UCAR Unidata Program
address@hidden                      http://my.unidata.ucar.edu