Newsletter #27

October 23, 1997

To subscribe/unsubscribe to the hdfnews mailing list, please send your
request to ncsalist@ncsa.uiuc.edu with the appropriate command (e.g.
subscribe hdfnews, unsubscribe hdfnews, help) in the *body* of the message.

CONTENTS

• HDF5 Prototype Release
• HDF 4.1r1 for T3E
• HDF Tutorial

The HDF5 Prototype release is now available, and can be downloaded from the HDF ftp server (hdf.ncsa.uiuc.edu):
https://support.hdfgroup.org/pub/outgoing/koziol/hdf5.tar.gz

The documentation is included with the release, and is also accessible from the URL:
http://hdf.ncsa.uiuc.edu/nra/BigHDF/

This release is intended primarily as a proof of concept and as a method of familiarizing users with the intended functionality.

When complete, HDF5 will include the following features:

• A new file format designed to address some of the deficiencies of the current version of HDF, particularly the need to store larger files and more objects per file.
• A simpler, more comprehensive data model that includes only two basic structures: a multi-dimensional array of record structures, and a grouping structure.
• A simpler, better-engineered library, with improved support for parallel I/O, threads, and other requirements imposed by modern systems and applications.
• A new API.

Limitations of the current prototype are:

• Multiple datasets may be created/read/written in an HDF5 file, but access must be as an entire dataset, i.e. no slab/slice/subset code is implemented yet. The datasets must also be homogeneous, orthogonal datasets, similar to the datasets which can be created in the HDF4 "SD" interface. Reducing these restrictions will be a major effort of further development on the library.
• Datasets can be located in a hierarchy of groups in the file, however user-level features for transitioning through the groups are not implemented in this release. Each dataset must be accessed through its full "pathname" in the file, ie. "/foo/bar/dataset1", "/foo/bar/dataset2", "/foo/data"

HDF5 has been tested on the following Unix platforms:

Linux, FreeBSD, IRIX, Solaris and Digital Unix

Machines which do not have IEEE floating-point representation or non big- or little-endian memory representations are not supported in this release. Most other machines should work correctly.

We have made changes to the HDF 4.1r1 release to support the T3E platform. This software did not go through a full QA cycle and is considered to be beta quality.

The netCDF (ncxxx) interfaces and ncdump, ncgen, and hdp do not work properly in this T3E release.

The source code and pre-compiled HDF 4.1r1 software for the T3E can be found on the NCSA ftp server (ftp.ncsa.uiuc.edu) in the directory: https://support.hdfgroup.org/HDF/HDF_Current/bin/t3e/

If you encounter any problems, please contact the HDF Helpdesk (hdfhelp@ncsa.uiuc.edu).

We have made a first cut of an HDF Tutorial. It includes the SD, Vdata, Vgroup, multi-file annotations (AN), and GR interfaces.

You can access it from the HDF home page under:
"Information about HDF -> Tutorials, Lectures, Examples ..."

(http://hdf.ncsa.uiuc.edu/tutslects.html)