HDF5 version 1.10.6 released on 2019-12-23
This document describes the differences between this release and the previous
HDF5 release. It contains information on the platforms tested and known
problems in this release. For more details check the HISTORY*.txt files in the
Note that documentation in the links below will be updated at the time of each
Links to HDF5 documentation can be found on The HDF5 web page:
The official HDF5 releases can be obtained from:
Changes from Release to Release and New Features in the HDF5-1.10.x release series
can be found at:
If you have any questions or comments, please send them to the HDF Help Desk:
- New Features
- Support for new platforms and languages
- Bug Fixes since HDF5-1.10.5
- Supported Platforms
- Tested Configuration Features Summary
- More Tested Platforms
- Known Problems
- CMake vs. Autotools installations
- Update CMake for VS2019 support
CMake added support for VS2019 in version 3.15. Changes to the CMake
generator setting required changes to scripts. Also updated version
references in CMake files as necessary.
(ADB - 2019/11/18, HDFFV-10962)
- Update CMake options to match new autotools options
Add configure options (autotools - CMake):
In addition NDEBUG is no longer forced defined and relies on the CMake
(ADB - 2019/10/07, HDFFV-100901, HDFFV-10637, TRILAB-97)
- Update CMake tests to use FIXTURES
CMake test fixtures allow setup/cleanup tests and other dependency
requirements as properties for tests. This is more flexible for
modern CMake code.
(ADB - 2019/07/23, HDFFV-10529)
- Windows PDB files are always installed
There are build configuration or flag settings for Windows that may not
generate PDB files. If those files are not generated then the install
utility will fail because those PDB files are not found. An optional
variable, DISABLE_PDB_FILES, was added to not install PDB files.
(ADB - 2019/07/17, HDFFV-10424)
- Add mingw CMake support with a toolchain file
There have been a number of mingw issues that have been linked under
HDFFV-10845. It has been decided to implement the CMake cross-compiling
technique of toolchain files. We will use a linux platform with the mingw
compiler stack for testing. Only the C language is fully supported, and
the error tests are skipped. The C++ language works for static but shared
builds have a shared library issue with the mingw Standard Exception Handling
library, which is not available on Windows. Fortran has a common cross-compile
problem with the fortran configure tests.
(ADB - 2019/07/12, HDFFV-10845, HDFFV-10595)
- Windows PDB files are installed incorrectly
For static builds, the PDB files for windows should be installed next
to the static libraries in the lib folder. Also the debug versions of
libraries and PDB files are now correctly built using the default
(ADB - 2019/07/09, HDFFV-10581)
- Add option to build only shared libs
A request was made to prevent building static libraries and only build
shared. A new option was added to CMake, ONLY_SHARED_LIBS, which will
skip building static libraries. Certain utility functions will build with
static libs but are not published. Tests are adjusted to use the correct
libraries depending on SHARED/STATIC settings.
(ADB - 2019/06/12, HDFFV-10805)
- Add options to enable or disable building tools and tests
Configure options --enable-tests and --enable-tools were added for
autotools configure. These options are enabled by default, and can be
disabled with either --disable-tests (or tools) or --enable-tests=no
(or --enable-tools=no). Build time is reduced ~20% when tools are
disabled, 35% when tests are disabled, 45% when both are disabled.
Re-enabling them after the initial build requires running configure
again with the option(s) enabled.
(LRK - 2019/06/12, HDFFV-9976)
- Change tools tests to search the error stack
There are some use cases which can cause the error stack of tools to be
different then the expected output. These tests now use grepTest.cmake;
this was changed to allow the error file to be searched for an expected string.
(ADB - 2019/04/15, HDFFV-10741)
- Added S3 and HDFS Virtual File Drivers (VFDs) to HDF5
These new VFDs have been introduced in HDF5-1.10.6. Instructions to
enable them when configuring HDF5 on Linux and Mac may be found at
Installing on Windows requires CMake 3.13 and the following additional setup.
Install openssl library (with dev files);
from "Shining Light Productions". msi package preferred.
PATH should have been updated with the installation dir.
set ENV variable OPENSSL_ROOT_DIR to the installation dir.
set ENV variable OPENSSL_CONF to the cfg file, likely %OPENSSL_ROOT_DIR%\bin\openssl.cfg
Install libcurl library (with dev files);
download the latest released version using git: https://github.com/curl/curl.git
Open a Visual Studio Command prompt
change to the libcurl root folder
run the "buildconf.bat" batch file
change to the winbuild directory
nmake /f Makefile.vc mode=dll MACHINE=x64
copy libcurl-vc-x64-release-dll-ipv6-sspi-winssl dir to C:\curl (installation dir)
set ENV variable CURL_ROOT to C:\curl (installation dir)
update PATH ENV variable to %CURL_ROOT%\bin (installation bin dir).
the aws credentials file should be in %USERPROFILE%\.aws folder
set the ENV variable "HDF5_ROS3_TEST_BUCKET_URL=https://s3.us-east-2.amazonaws.com/hdf5ros3"
(ADB - 2019/09/12, HDFFV-10854)
- Added new wrappers for H5Pset/get_create_intermediate_group()
(BMR - 2019/04/22, HDFFV-10622)
- Fixed a failure in JUnit-TestH5P on 32-bit architectures
(JTH - 2019/04/30)
Support for new platforms, languages and compilers.
- CMake added support for VS2019 in version 3.15. Updated scripts.
- macOS 10.13.6 Darwin 17.7.0 with Apple clang LLVM version 10.0.0
- macOS 10.14.6 Darwin 18.7.0 with Apple clang LLVM version 10.0.1
Bug Fixes since HDF5-1.10.5 release
- Improved performance when creating a large number of small datasets by
retrieving default property values from the API context instead of doing
skip list searches. More work is required to achieve parity with HDF5 1.8.
(CJH - 2019/12/10, HDFFV-10658)
- Fixed user-created data access properties not existing in the property list
returned by H5Dget_access_plist. Thanks to Steven Varga for submitting a
reproducer and a patch.
(CJH - 2019/12/9, HDFFV-10934)
- Inappropriate linking with deprecated MPI C++ libraries
HDF5 does not define *_SKIP_MPICXX in the public headers, so applications
can inadvertently wind up linking to the deprecated MPI C++ wrappers.
MPICH_SKIP_MPICXX and OMPI_SKIP_MPICXX have both been defined in H5public.h
so this should no longer be an issue. HDF5 makes no use of the deprecated
MPI C++ wrappers.
(DER - 2019/09/17, HDFFV-10893)
- fcntl(2)-based file locking incorrectly passed the lock argument struct
instead of a pointer to the struct, causing errors on systems where
flock(2) is not available.
File locking is used when files are opened to enforce SWMR semantics. A
lock operation takes place on all file opens unless the
HDF5_USE_FILE_LOCKING environment variable is set to the string "FALSE".
flock(2) is preferentially used, with fcntl(2) locks as a backup if
flock(2) is unavailable on a system (if neither is available, the lock
operation fails). On these systems, the file lock will often fail, which
causes HDF5 to not open the file and report an error.
This bug only affects POSIX systems. Win32 builds on Windows use a no-op
locking call which always succeeds. Systems which exhibit this bug will
have H5_HAVE_FCNTL defined but not H5_HAVE_FLOCK in the configure output.
This bug affects HDF5 1.10.0 through 1.10.5.
fcntl(2)-based file locking now correctly passes the struct pointer.
(DER - 2019/08/27, HDFFV-10892)
- Fixed a bug caused by a bad tag value when condensing object header
There was an assertion failure when moving messages from running a
user test program with library release HDF5 1.10.4. It was because
the tag value (object header's address) was not set up when entering
the library routine H5O__chunk_update_idx(), which eventually
verifies the metadata tag value when protecting the object header.
The problem was fixed by replacing FUNC_ENTER_PACKAGE in H5O__chunk_update_idx()
with FUNC_ENTER_PACKAGE_TAG(oh->cache_info.addr) to set up the metadata tag.
(VC - 2019/08/23, HDFFV-10873)
- Fixed the test failure from test_metadata_read_retry_info() in
The test failure is due to an incorrect number of bins returned for
retry info (info.nbins). The # of bins expected for 101 read attempts
is 3 instead of 2. The routine H5F_set_retries() in src/H5Fint.c
calculates the # of bins by first obtaining the log10 value for
(read attempts - 1). For PGI/19, the log10 value for 100 read attempts
is 1.9999999999999998 instead of 2.00000. When casting the log10 value
to unsigned later on, the decimal part is chopped off causing the test
This was fixed by obtaining the rounded integer value (HDceil) for the
log10 value of read attempts first before casting the result to unsigned.
(VC - 2019/8/14, HDFFV-10813)
- Fixed an issue when creating a file with non-default file space info
together with library high bound setting to H5F_LIBVER_V18.
When setting non-default file space info in fcpl via
H5Pset_file_space_strategy() and then creating a file with both high and
low library bounds set to H5F_LIBVER_V18 in fapl, the library succeeds in
creating the file. File creation should fail because the feature of
setting non-default file space info does not exist in library release 1.8
This was fixed by setting and checking the proper version in the file
space info message based on the library low and high bounds when creating
and opening the HDF5 file.
(VC - 2019/6/25, HDFFV-10808)
- Fixed an issue where copying a version 1.8 dataset between files using
H5Ocopy fails due to an incompatible fill version
When using the HDF5 1.10.x H5Ocopy() API call to copy a version 1.8
dataset to a file created with both high and low library bounds set to
H5F_LIBVER_V18, the H5Ocopy() call will fail with the error stack indicating
that the fill value version is out of bounds.
This was fixed by changing the fill value message version to H5O_FILL_VERSION_3
(from H5O_FILL_VERSION_2) for H5F_LIBVER_V18.
(VC - 2019/6/14, HDFFV-10800)
- Fixed a bug that would cause an error or cause fill values to be
incorrectly read from a chunked dataset using the "single chunk" index if
the data was held in cache and there was no data on disk.
(NAF - 2019/03/06)
- Fixed a bug that could cause an error or cause fill values to be
incorrectly read from a dataset that was written to using H5Dwrite_chunk
if the dataset was not closed after writing.
(NAF - 2019/03/06, HDFFV-10716)
- Fixed memory leak in scale offset filter
In a special case where the MinBits is the same as the number of bits in
the datatype's precision, the filter's data buffer was not freed, causing
the memory usage to grow. In general the buffer was freed correctly. The
Minbits are the minimal number of bits to store the data values. Please
see the reference manual for H5Pset_scaleoffset for the details.
(RL - 2019/3/4, HDFFV-10705)
- Correct option for default API version
CMake options for default API version are not mutually exclusive.
Change the multiple BOOL options to a single STRING option with the
strings; v16, v18, v110.
(ADB - 2019/08/12, HDFFV-10879)
- h5repack was fixed to repack datasets with external storage
to other types of storage.
New test added to repack files and verify the correct data using h5diff.
(JS - 2019/09/25, HDFFV-10408)
(ADB - 2019/10/02, HDFFV-10918)
Linux 2.6.32-696.20.1.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23)
IBM XL C/C++ V13.1
IBM XL Fortran V15.1
Linux 3.10.0-327.10.1.el7 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(jelly/kituo/moohan) Version 4.8.5 20150623 (Red Hat 4.8.5-4)
Version 4.9.3, Version 5.2.0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
Version 17.0.0.098 Build 20160721
MPICH 3.1.4 compiled with GCC 4.9.3
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2015 w/ Intel Fortran 18 (cmake)
Windows 7 x64 Visual Studio 2015 w/ Intel C, Fortran 2018 (cmake)
Visual Studio 2015 w/ MSMPI 8 (cmake)
Windows 10 Visual Studio 2015 w/ Intel Fortran 18 (cmake)
Windows 10 x64 Visual Studio 2015 w/ Intel Fortran 18 (cmake)
Visual Studio 2017 w/ Intel Fortran 19 (cmake)
Visual Studio 2019 w/ Intel Fortran 19 (cmake)
macOS 10.13.6, Darwin, Apple clang LLVM version 10.0.0
17.7.0, x86_64 gfortran GNU Fortran (GCC) 6.3.0
(bear) Intel icc/icpc/ifort version 19.0.4
macOS 10.14.6, Darwin Apple clang LLVM version 10.0.1
18.7.0, x86_64 gfortran GNU Fortran (GCC) 6.3.0
(bobcat) Intel icc/icpc/ifort version 19.0.4
Tested Configuration Features Summary
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.11 32-bit n y/y n y y y
Solaris2.11 64-bit n y/n n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y y y y y
Windows 7 Cygwin n y/n n y y y
Windows 7 x64 Cygwin n y/n n y y y
Windows 10 y y/y n y y y
Windows 10 x64 y y/y n y y y
Mac OS X Yosemite 10.10.5 64-bit n y/y n y y y
Mac OS X El Capitan 10.11.6 64-bit n y/y n y y y
MacOS High Sierra 10.13.6 64-bit n y/y n y y y
CentOS 7.2 Linux 3.10.0 x86_64 PGI n y/y n y y y
CentOS 7.2 Linux 3.10.0 x86_64 GNU y y/y y y y y
CentOS 7.2 Linux 3.10.0 x86_64 Intel n y/y n y y y
Linux 2.6.32-573.18.1.el6.ppc64 n y/y n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.11 32-bit y y y y
Solaris2.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 7 x64 Cygwin n n n y
Windows 10 y y y y
Windows 10 x64 y y y y
Mac OS X Yosemite 10.10.5 64-bit y y y y
Mac OS X El Capitan 10.11.6 64-bit y y y y
MacOS High Sierra 10.13.6 64-bit y y y y
CentOS 7.2 Linux 3.10.0 x86_64 PGI y y y n
CentOS 7.2 Linux 3.10.0 x86_64 GNU y y y y
CentOS 7.2 Linux 3.10.0 x86_64 Intel y y y n
Linux 2.6.32-573.18.1.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
The following configurations are not supported but have been tested for this release.
Linux 2.6.32-754.11.1.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(mayll/platypus) Version 4.4.7 20120313
Version 4.9.3, 5.3.0, 6.2.0
PGI C, Fortran, C++ for 64-bit target on
Intel(R) C (icc), C++ (icpc), Fortran (icc)
Version 220.127.116.11 Build 20170411
MPICH 3.1.4 compiled with GCC 4.9.3
Linux 3.10.0-327.18.2.el7 GNU C (gcc) and C++ (g++) compilers
#1 SMP x86_64 GNU/Linux Version 4.8.5 20150623 (Red Hat 4.8.5-4)
(jelly) with NAG Fortran Compiler Release 6.2(Chiyoda)
GCC Version 7.1.0
Intel(R) C (icc) and C++ (icpc) compilers
Version 17.0.0.098 Build 20160721
with NAG Fortran Compiler Release 6.2(Chiyoda)
Linux 3.10.0-327.10.1.el7 MPICH 3.2 compiled with GCC 5.3.0
#1 SMP x86_64 GNU/Linux
Linux 2.6.32-573.18.1.el6.ppc64 MPICH mpich 3.1.4 compiled with
#1 SMP ppc64 GNU/Linux IBM XL C/C++ for Linux, V13.1
(ostrich) and IBM XL Fortran for Linux, V15.1
#1 SMP x86_64 GNU/Linux GNU gcc (GCC) 9.2.1 20190827 (Red Hat 9.2.1 20190827)
GNU Fortran (GCC) 9.2.1 20190827 (Red Hat 9.2.1 20190827)
(cmake and autotools)
Mac OS X 10.11.6, Darwin, Apple clang version 7.3.0 from Xcode 7.3
15.6.0, x86-64 gfortran GNU Fortran (GCC) 5.2.0
(osx1011test) Intel icc/icpc/ifort version 16.0.2
macOS 10.12.6, Darwin, Apple clang LLVM version 8.1.0 from Xcode 8.3
16.6.0, x86_64 gfortran GNU Fortran (GCC) 7.1.0
(kite) Intel icc/icpc/ifort version 17.0.2
Windows 7 x64 Visual Studio 2008
CMake files do not behave correctly with paths containing spaces.
Do not use spaces in paths because the required escaping for handling spaces
results in very complex and fragile build files.
ADB - 2019/05/07
At present, metadata cache images may not be generated by parallel
applications. Parallel applications can read files with metadata cache
images, but since this is a collective operation, a deadlock is possible
if one or more processes do not participate.
Three tests fail with OpenMPI 3.0.0/GCC-7.2.0-2.29:
testphdf5 (ecdsetw, selnone, cchunk1, cchunk3, cchunk4, and actualio)
t_pflush1/fails on exit
The first two tests fail attempting collective writes.
Parallel builds using OpenMPI 3.03 or later and romio fail several tests
with collective writes or compression that will not fail when ompio is used
instead of romio. This can be done by adding "--mca io ompio" to the mpirun
command. For example, in autotools builds RUNPARALLEL can be set to
"mpirun --mca io ompio -n 6" provided ompio is installed.
CPP ptable test fails on VS2017 with Intel compiler, JIRA issue: HDFFV-10628.
This test will pass with VS2015 with Intel compiler.
Older MPI libraries such as OpenMPI 2.0.1 and MPICH 2.1.5 were tested
while attempting to resolve the Jira issue: HDFFV-10540.
The known problems of reading or writing > 2GBs when using MPI-2 was
partially resolved with the MPICH library. The proposed support recognizes
IO operations > 2GB and if the datatype is not a derived type, the library
breaks the IO into chunks which can be input or output with the existing
MPI 2 limitations, i.e. size reporting and function API size/count
arguments are restricted to be 32 bit integers. For derived types larger
than 2GB, MPICH 2.1.5 fails while attempting to read or write data.
OpenMPI in contrast, implements MPI-3 APIs even in the older releases
and thus does not suffer from the 32 bit size limitation described here.
OpenMPI releases prior to v3.1.3 appear to have other datatype issues however,
e.g. within a single parallel test (testphdf5) the subtests (cdsetr, eidsetr)
report data verfication errors before eventually aborting.
The most recent versions of OpenMPI (v3.1.3 or newer) have evidently
resolved these isses and parallel HDF5 testing does not currently report
errors though occasional hangs have been observed.
Known problems in previous releases can be found in the HISTORY*.txt files
in the HDF5 source. Please report any new problems found to
CMake vs. Autotools installations
While both build systems produce similar results, there are differences.
Each system produces the same set of folders on linux (only CMake works
on standard Windows); bin, include, lib and share. Autotools places the
COPYING and RELEASE.txt file in the root folder, CMake places them in
the share folder.
The bin folder contains the tools and the build scripts. Additionally, CMake
creates dynamic versions of the tools with the suffix "-shared". Autotools
installs one set of tools depending on the "--enable-shared" configuration
Autotools: h5c++, h5cc, h5fc
CMake: h5c++, h5cc, h5hlc++, h5hlcc
The include folder holds the header files and the fortran mod files. CMake
places the fortran mod files into separate shared and static subfolders,
while Autotools places one set of mod files into the include folder. Because
CMake produces a tools library, the header files for tools will appear in
the include folder.
The lib folder contains the library files, and CMake adds the pkgconfig
subfolder with the hdf5*.pc files used by the bin/build scripts created by
the CMake build. CMake separates the C interface code from the fortran code by
creating C-stub libraries for each Fortran library. In addition, only CMake
installs the tools library. The names of the szip libraries are different
between the build systems.
The share folder will have the most differences because CMake builds include
a number of CMake specific files for support of CMake's find_package and support
for the HDF5 Examples CMake project.