Rhdf5lib 1.24.2
Rhdf5lib provides versions of the C and C++ HDF5 libraries. It is primarily useful to developers of other R packages who want to make use of the capabilities of the HDF5 library directly in the C or C++ code of their own packages, rather than using a higher level interface such as the rhdf5 package. Using Rhdf5lib makes life easier for users, as they do not have to worry about installing libraries at a system level, and for developers since they can work with a defined version of the library rather than developing strategies to cope with the potential for multiple versions.
Rhdf5lib is very much inspired by the zlibbioc and Rhtslib packages.
There is an example package, usingRhdf5lib, that demonstrates how packages should link to Rhdf5lib.
To link successfully to the HDF5 library included in Rhdf5lib a package must include both a src/Makevars.win
and src/Makevars
file.
Add the following lines to both src/Makevars
and src/Makevars.win
RHDF5_LIBS=$(shell "${R_HOME}/bin${R_ARCH_BIN}/Rscript" -e \
'Rhdf5lib::pkgconfig("PKG_C_LIBS")')
PKG_LIBS=$(RHDF5_LIBS)
The statement for each platform modifies the $PKG_LIBS
variable. If your package needs to add additional information to the $PKG_LIBS
variable, do so by adding to the PKG_LIBS=$(RHDF5_LIBS)
line, e.g.,
PKG_LIBS=$(RHDF5_LIBS) -L/path/to/foolib -lfoo
Note that the use of $(shell ...)
necessitates using GNU Make, and you need to make this requirement explicit in your package’s DESCRIPTION file via the entry:
SystemRequirements: GNU make
The default behaviour of Rhdf5lib::pkgconfig
is to report the location of the shared library as the result of system.file("lib", package="Rhdf5lib")
. If this is inappropriate for your system e.g. a cluster with a shared file system, use the environment variable RHDF5LIB_RPATH
to override this and set an appropriate location for your infrastructure.
Valid options to provide to pkgconfig()
are: PKG_C_LIBS
, PKG_CXX_LIBS
, PKG_C_HL_LIBS
and PKG_CXX_HL_LIBS
. Choose the most appropriate depending upon whether your linking code requires the C++ API (C
vs CXX
) and/or the HDF5 ‘high-level’ API (HL
). Choosing options that you don’t require should not harm performance, but will result in a larger library and potentially greater memory usage for your application, so it is good practice to select only the features you need.
In order for the C/C++ compiler to find the HDF5 headers during package installation, add Rhdf5lib to the LinkingTo field of the DESCRIPTION file of your package, e.g.
LinkingTo: Rhdf5lib
In you C or C++ code files, you can then use the standard include techniques, e.g., #include "hdf5.h"
or #include "H5Cpp.h"
. You can inspect the header files manually to check their names and declared functions. To find their location on your system you can use the following code:
system.file(package="Rhdf5lib", "include")
## [1] "/tmp/RtmpAoKNVo/Rinst1ac050529d7d69/Rhdf5lib/include"
Rhdf5lib requires the ZLIB compression library to be installed on non-Windows platforms. If installation fails with a message reporting that zlib.h can not be found, it is possible to provide the appropriate path explicitly during installation via the configure.args
argument e.g.
BiocManager::install('Rhdf5lib', configure.args = "--with-zlib='/path/to/zlib/'")
Here /path/to/zlib
should be the directory that contains both include/zlib.h
and lib/libz.a
. For example, on a typical Ubuntu installation this may be /usr/
while for libraries installed via miniconda this location could be /home/<USER>/miniconda3/
.
If you have configured R with link-time optimisation enabled (see here), but wish to turn it off for the installation of Rhdf5lib, you need to set both the configure.args
and INSTALL_opts
arguments e.g.
BiocManager::install('Rhdf5lib', INSTALL_opts="--no-use-LTO", configure.args = "--disable-lto")
This is because building Rhdf5lib involves first compiling the HDF5 library, and then compiling and linking the R interface against that. The INSTALL_opts
argument affects the latter part, but we need to use configure.args
to ensure the HDF5 library is built with the same settings. 1 Using "--enable-lto"
here will have no effect. To enable link-time optimisation you must have already configured R with --enable-lto
(see here).
If you encounter problems checking whether to use -Wl,-rpath
to “link shared libs in nondefault directories” you can disable the test by passing the option "--disable-sharedlib-rath"
to the configuration script.
BiocManager::install('Rhdf5lib', configure.args = "--disable-sharedlib-rath")
MLS was supported by the BMBF-funded Heidelberg Center for Human Bioinformatics (HD-HuB) within the German Network for Bioinformatics Infrastructure (de.NBI), Grant Number #031A537B
## R version 4.3.2 Patched (2023-11-13 r85521)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 22.04.3 LTS
##
## Matrix products: default
## BLAS: /home/biocbuild/bbs-3.18-bioc/R/lib/libRblas.so
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## [3] LC_TIME=en_GB LC_COLLATE=C
## [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
## [9] LC_ADDRESS=C LC_TELEPHONE=C
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: America/New_York
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] BiocStyle_2.30.0
##
## loaded via a namespace (and not attached):
## [1] digest_0.6.34 R6_2.5.1 bookdown_0.37
## [4] fastmap_1.1.1 xfun_0.41 cachem_1.0.8
## [7] knitr_1.45 htmltools_0.5.7 rmarkdown_2.25
## [10] lifecycle_1.0.4 cli_3.6.2 sass_0.4.8
## [13] jquerylib_0.1.4 compiler_4.3.2 tools_4.3.2
## [16] evaluate_0.23 bslib_0.6.1 yaml_2.3.8
## [19] BiocManager_1.30.22 jsonlite_1.8.8 rlang_1.1.3