Merge pull request #110 from TREX-CoE/no_hl

Removed dependency on hdf5_hl library

ChangeLog

@@ -4,6 +4,7 @@ CHANGES
 2.3
 ---
 
+- Removed dependency to hdf5_hl
 - Fixed Fortran interface on 32-bit (e.g. i386) architectures
 - Changed the structure of the state group
 - Sparse data type is patched to work with different dimensions

README.md

@@ -5,7 +5,7 @@
 [![build](https://github.com/TREX-CoE/trexio/actions/workflows/actions.yml/badge.svg)](https://github.com/TREX-CoE/trexio/actions/workflows/actions.yml)
 ![GitHub release (latest by date)](https://img.shields.io/github/v/release/TREX-CoE/trexio)
 
-TREX library for efficient I/O.
+TREX library for efficient I/O. 
 
 
 ## Minimal requirements (for users):
@@ -37,6 +37,8 @@ TREX library for efficient I/O.
 - Emacs         (>= 26.0)
 - SWIG          (>= 4.0)   [required for the Python API]
 
+**Note:** The source code is auto-generated from the Emacs org-mode (`.org`) files following the literate programming approach. This is why the `src` directory is initially empty.
+
 ## Installation procedure from the GitHub repo clone (for developers):
 
 1. `git clone https://github.com/TREX-CoE/trexio.git`
@@ -83,13 +85,10 @@ The official releases of TREXIO `>=2.0.0` can be installed using the
 [GNU Guix](https://guix.gnu.org) functional package manager.
 The [trexio.scm](https://github.com/TREX-CoE/trexio/blob/master/tools/trexio.scm)
 Schema file contains the manifest specification for the `trexio` package.
-It can be installed within the selected `$GUIX_PROFILE` as follows:
+It can be installed as follows:
 
 ```
-guix package \
+guix package --cores=<n_cores> --install-from-file=trexio.scm
-	--profile=$GUIX_PROFILE \
-	--cores=<n_cores> \
-	--install-from-file=trexio.scm
 ```
 
 ## Installation procedure for Spack users
@@ -104,6 +103,18 @@ It can be installed as follows
 spack install --jobs <n_cores> trexio
 ```
 
+## Installation procedure for Debian/Ubuntu users
+
+The official release of TREXIO `2.2.0` is available as a Debian (`.deb`) package thanks to the [Debichem Team](https://wiki.debian.org/Debichem).
+The source code is hosted [here](https://salsa.debian.org/debichem-team/libtrexio) and 
+the pre-built binary files are available via the [Debian package registry](https://packages.debian.org/bookworm/libtrexio-dev).
+
+TREXIO is also available on [Ubuntu 23.04 (Lunar Lobster)](https://packages.ubuntu.com/lunar/libtrexio-dev) and newer and can be installed as follows:
+
+```
+sudo apt-get update && sudo apt-get install libtrexio-dev
+```
+
 ## Compilation without the HDF5 library
 
 By default, the configuration step proceeds to search for the [HDF5 library](https://portal.hdfgroup.org/display/HDF5/HDF5).

configure.ac

@@ -95,7 +95,7 @@ AC_RUN_IFELSE(
 execl("/bin/sh", "sh", "-c", "mkdir tmpdir1 && \
 touch tmpdir1/test_file && \
 $CP_PROG -r -n tmpdir1 tmpdir2 && \
-exec ls tmpdir2/test_file", NULL);
+exec ls tmpdir2/test_file > /dev/null", NULL);
 ])],
 [ rm -rf tmpdir1 tmpdir2
   CP_COMMAND="\"$CP_PROG\", \"-r\", \"-n\""
@@ -122,7 +122,9 @@ AC_CHECK_HEADERS([fcntl.h inttypes.h stdint.h stdbool.h stdlib.h string.h unistd
 ### ----
 
 HDF5_LIBS=""
+HDF5_LDFLAGS=""
 HDF5_CFLAGS=""
+HDF5_CPPFLAGS=""
 AC_ARG_WITH([hdf5],
   AS_HELP_STRING([--with-hdf5=PATH], [Path to HDF5 library and headers]), [
   with_hdf5="$withval"], [with_hdf5="yes"])
@@ -130,29 +132,87 @@ AC_ARG_WITH([hdf5],
 AS_IF([test "x$with_hdf5" == "xno"], [
            AC_DEFINE([HAVE_HDF5], 0, [Define to 1 if HDF5 is available]) ],
       [test "x$with_hdf5" != "xyes"], [
+           HDF5_LIBS="-lhdf5"
            HDF5_PATH="$with_hdf5"
-           HDF5_LIBS="-L$HDF5_PATH/lib -lhdf5 -lhdf5_hl"
+           HDF5_LDFLAGS="-L$HDF5_PATH/lib"
-           HDF5_CFLAGS="-I$HDF5_PATH/include"
+           HDF5_CPPFLAGS="-I$HDF5_PATH/include"
            AC_DEFINE([HAVE_HDF5], 1, [Define to 1 if HDF5 is available]) ],
       [
-           PKG_CHECK_MODULES([HDF5], [hdf5 >= 1.8])
+           PKG_CHECK_EXISTS([hdf5], [
-           AS_IF([ test "${HDF5_LIBS}" == "" ], [
+             PKG_CHECK_MODULES([HDF5], [hdf5 >= 1.8])
-             AC_CHECK_LIB([hdf5], [H5Fcreate], [ HDF5_LIBS="-lhdf5" ])
+           ],
-             AC_CHECK_HEADER([hdf5.h], [HDF5_CFLAGS=""], [AC_MSG_ERROR([HDF5 header not found]) ])
+           [ AC_PATH_PROG([H5CC],[h5cc],[not_found])
-             AC_CHECK_HEADER([hdf5_hl.h], [HDF5_CFLAGS=""], [AC_MSG_ERROR([HDF5 header not found]) ])
+             AS_IF([test "$H5CC" != "not_found"], [
+              HDF5_LIBS="-lhdf5"
+              AC_REQUIRE([AC_PROG_SED])
+              AC_REQUIRE([AC_PROG_AWK])
+              AC_REQUIRE([AC_PROG_GREP])
+              # Look for "HDF5 Version: X.Y.Z"
+              HDF5_VERSION=$(eval $H5CC -showconfig | $GREP 'HDF5 Version:' \
+                  | $AWK '{print $[]3}')
+
+              # A ideal situation would be where everything we needed was
+              # in the AM_* variables. However most systems are not like this
+              # and seem to have the values in the non-AM variables.
+              #
+              # We try the following to find the flags:
+              # (1) Look for "NAME:" tags
+              # (2) Look for "H5_NAME:" tags
+              # (3) Look for "AM_NAME:" tags
+              #
+              HDF5_tmp_flags=$(eval $H5CC -showconfig \
+                  | $GREP 'FLAGS\|Extra libraries:' \
+                  | $AWK -F: '{printf("%s "), $[]2}' )
+
+              dnl Find the installation directory and append include/
+              HDF5_tmp_inst=$(eval $H5CC -showconfig \
+                  | $GREP 'Installation point:' \
+                  | $AWK '{print $[]NF}' )
+
+              dnl Add this to the CPPFLAGS
+              HDF5_CPPFLAGS="-I${HDF5_tmp_inst}/include"
+
+              HDF5_SHOW=$(eval $H5CC -show)
+
+              dnl Now sort the flags out based upon their prefixes
+              for arg in $HDF5_SHOW $HDF5_tmp_flags ; do
+                case "$arg" in
+                  -I*) echo $HDF5_CPPFLAGS | $GREP -e "$arg" 2>&1 >/dev/null \
+                        || HDF5_CPPFLAGS="$HDF5_CPPFLAGS $arg"
+                    ;;
+                  -L*) echo $HDF5_LDFLAGS | $GREP -e "$arg" 2>&1 >/dev/null \
+                        || HDF5_LDFLAGS="$HDF5_LDFLAGS $arg"
+                    ;;
+                  -l*) echo $HDF5_LIBS | $GREP -e "$arg" 2>&1 >/dev/null \
+                        || HDF5_LIBS="$HDF5_LIBS $arg"
+                    ;;
+                esac
+              done
+
+             ])
            ])
            AC_DEFINE([HAVE_HDF5], 1, [Define to 1 if HDF5 is available])
-           HDF5_LIBS="$HDF5_LIBS -lhdf5_hl"
       ])
 
 AM_CONDITIONAL([HAVE_HDF5],[test "x$with_hdf5" != "xno"])
 
 AC_SUBST([HDF5_LDFLAGS])
+AC_SUBST([HDF5_LIBS])
 AC_SUBST([HDF5_CFLAGS])
 AC_SUBST([HDF5_CPPFLAGS])
+CPPFLAGS="${HDF5_CPPFLAGS} ${CPPFLAGS}"
 CFLAGS="${HDF5_CFLAGS} ${CFLAGS}"
+LDFLAGS="${HDF5_LDFLAGS} ${LDFLAGS}"
 LIBS="${HDF5_LIBS} ${LIBS}"
 
+AS_IF([test "x$with_hdf5" != "xno"], [
+  OLD_LIBS=$LIBS
+  AC_CHECK_LIB([hdf5], [H5Fcreate], [], [
+    AC_MSG_ERROR([-lhdf5 fails, use ./configure --with-hdf5=...]) ])
+  LIBS=$OLD_LIBS
+  AC_CHECK_HEADER([hdf5.h], [], [
+    AC_MSG_ERROR([hdf5.h not found, use ./configure --with-hdf5=...]) ])
+])
 
 # The block below should only execute if the ax_lib_hdf5.m4 macro failed to find HDF5.
 # It is only needed to manually build Python API because setup.py depends on HDF5.

python/setup.py

@@ -92,7 +92,7 @@ h5_ldflags = h5_ldflags_withl.split(" ")[0] if not do_sdist else ""
 pytrexio_module = Extension('pytrexio._pytrexio',
                             sources = [os.path.join(srcpath, code) for code in c_files],
                             include_dirs = [h5_cflags, srcpath, numpy_includedir],
-                            libraries = ['hdf5', 'hdf5_hl'],
+                            libraries = ['hdf5' ],
                             extra_compile_args = [
                                 '-std=c99',
                                 '-Wno-discarded-qualifiers',

src/templates_hdf5/templator_hdf5.org

@@ -39,7 +39,6 @@
 #include <sys/stat.h>
 
 #include "hdf5.h"
-#include "hdf5_hl.h"
 
   #+end_src
 
@@ -380,18 +379,21 @@ trexio_hdf5_read_$group_dset$ (trexio_t* const file, $group_dset_dtype$* const $
   }
 
   FREE(ddims);
+
+ /* Read dataset */
+  herr_t status = H5Dread(dset_id,
+                          H5T_$GROUP_DSET_H5_DTYPE$,
+                          H5S_ALL, H5S_ALL, H5P_DEFAULT, 
+                          $group_dset$);
+
   H5Sclose(dspace_id);
   H5Dclose(dset_id);
 
- /* High-level H5LT API. No need to deal with dataspaces and datatypes */
-  herr_t status = H5LTread_dataset(f->$group$_group,
-                                   $GROUP_DSET$_NAME,
-                                   H5T_$GROUP_DSET_H5_DTYPE$,
-                                   $group_dset$);
   if (status < 0) return TREXIO_FAILURE;
 
   return TREXIO_SUCCESS;
 }
+
    #+end_src
 
    #+begin_src c :tangle write_dset_data_hdf5.c
@@ -411,7 +413,8 @@ trexio_hdf5_write_$group_dset$ (trexio_t* const file, const $group_dset_dtype$*
 
      Consider using HDF5-native h5repack utility after deleting/overwriting big datasets.
   ,*/
-  if (H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME) == 1 && file->mode == 'u') {
+  
+  if ((trexio_hdf5_has_$group_dset$(file) == TREXIO_SUCCESS) && (file->mode == 'u')) {
     herr_t status_del = H5Ldelete(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
     if (status_del < 0) return TREXIO_FAILURE;
   }
@@ -454,14 +457,13 @@ trexio_hdf5_has_$group_dset$ (trexio_t* const file)
   trexio_hdf5_t* f = (trexio_hdf5_t*) file;
   if (f->$group$_group == (hsize_t) 0) return TREXIO_HAS_NOT;
 
-  herr_t status = H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME);
+  htri_t exists = H5Lexists(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
- /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
+  if (exists > 0) {
-  if (status == 1){
     return TREXIO_SUCCESS;
-  } else if (status == 0) {
+  } else if (exists < 0) {
-    return TREXIO_HAS_NOT;
-  } else {
     return TREXIO_FAILURE;
+  } else {
+    return TREXIO_HAS_NOT;
   }
 
 }
@@ -534,7 +536,7 @@ trexio_hdf5_write_$group_dset$ (trexio_t* const file,
 
   trexio_exit_code rc_write = TREXIO_FAILURE;
  /* NOTE: chunk size is set upon creation of the HDF5 dataset and cannot be changed ! */
-  if ( H5LTfind_dataset(f->$group$_group, dset_index_name) != 1 ) {
+  if (trexio_hdf5_has_$group_dset$(file) == TREXIO_HAS_NOT) {
  /* If the file does not exist -> create it and write */
 
  /* Create chunked dataset with index_dtype datatype and write indices into it */
@@ -657,14 +659,13 @@ trexio_hdf5_has_$group_dset$ (trexio_t* const file)
   trexio_hdf5_t* f = (trexio_hdf5_t*) file;
   if (f->$group$_group == (hsize_t) 0) return TREXIO_HAS_NOT;
 
-  herr_t status = H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME "_values");
+  htri_t exists = H5Lexists(f->$group$_group, $GROUP_DSET$_NAME "_values", H5P_DEFAULT);
- /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
+  if (exists > 0) {
-  if (status == 1){
     return TREXIO_SUCCESS;
-  } else if (status == 0) {
+  } else if (exists < 0) {
-    return TREXIO_HAS_NOT;
-  } else {
     return TREXIO_FAILURE;
+  } else {
+    return TREXIO_HAS_NOT;
   }
 
 }
@@ -728,7 +729,7 @@ trexio_exit_code trexio_hdf5_write_$group_dset$(trexio_t* const file,
 
   trexio_exit_code rc_write = TREXIO_FAILURE;
  /* NOTE: chunk size is set upon creation of the HDF5 dataset and cannot be changed ! */
-  if ( H5LTfind_dataset(f->$group$_group, dset_name) != 1 ) {
+  if (trexio_hdf5_has_$group_dset$(file) == TREXIO_HAS_NOT) {
  /* If the file does not exist -> create it and write */
 
  /* Create chunked dataset with dtype datatype and write indices into it */
@@ -793,14 +794,13 @@ trexio_exit_code trexio_hdf5_has_$group_dset$(trexio_t* const file)
 
   const char dset_name[256] = "$group_dset$";
 
-  herr_t status = H5LTfind_dataset(f->$group$_group, dset_name);
+  htri_t exists = H5Lexists(f->$group$_group, dset_name, H5P_DEFAULT);
- /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
+  if (exists > 0) {
-  if (status == 1){
     return TREXIO_SUCCESS;
-  } else if (status == 0) {
+  } else if (exists < 0) {
-    return TREXIO_HAS_NOT;
-  } else {
     return TREXIO_FAILURE;
+  } else {
+    return TREXIO_HAS_NOT;
   }
 }
    #+end_src
@@ -936,7 +936,7 @@ trexio_hdf5_write_$group_dset$ (trexio_t* const file, const char** $group_dset$,
 
      Consider using HDF5-provided h5repack utility after deleting/overwriting big datasets.
   ,*/
-  if (H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME) == 1 && file->mode == 'u') {
+  if ( (trexio_hdf5_has_$group_dset$(file) == TREXIO_SUCCESS) && (file->mode == 'u') ) {
     herr_t status_del = H5Ldelete(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
     if (status_del < 0) return TREXIO_FAILURE;
   }
@@ -991,14 +991,13 @@ trexio_hdf5_has_$group_dset$ (trexio_t* const file)
   trexio_hdf5_t* f = (trexio_hdf5_t*) file;
   if (f->$group$_group == (hsize_t) 0) return TREXIO_HAS_NOT;
 
-  herr_t status = H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME);
+  htri_t exists = H5Lexists(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
- /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
+  if (exists > 0) {
-  if (status == 1){
     return TREXIO_SUCCESS;
-  } else if (status == 0) {
+  } else if (exists < 0) {
-    return TREXIO_HAS_NOT;
-  } else {
     return TREXIO_FAILURE;
+  } else {
+    return TREXIO_HAS_NOT;
   }
 
 }
@@ -1231,7 +1230,7 @@ trexio_exit_code trexio_hdf5_write_determinant_list(trexio_t* const file,
 
   trexio_exit_code rc_write = TREXIO_FAILURE;
  /* NOTE: chunk size is set upon creation of the HDF5 dataset and cannot be changed ! */
-  if ( H5LTfind_dataset(f->determinant_group, dset_det_name) != 1 ) {
+  if ( trexio_hdf5_has_determinant_list(file) == TREXIO_HAS_NOT ) {
  /* If the file does not exist -> create it and write */
 
  /* Create chunked dataset with det_dtype datatype and write indices into it */
@@ -1260,14 +1259,13 @@ trexio_exit_code trexio_hdf5_has_determinant_list(trexio_t* const file)
   trexio_hdf5_t* f = (trexio_hdf5_t*) file;
   if (f->determinant_group == (hsize_t) 0) return TREXIO_HAS_NOT;
 
-  herr_t status = H5LTfind_dataset(f->determinant_group, "determinant_list");
+  htri_t exists = H5Lexists(f->determinant_group, "determinant_list", H5P_DEFAULT);
- /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
+  if (exists > 0) {
-  if (status == 1){
     return TREXIO_SUCCESS;
-  } else if (status == 0) {
+  } else if (exists < 0) {
-    return TREXIO_HAS_NOT;
-  } else {
     return TREXIO_FAILURE;
+  } else {
+    return TREXIO_HAS_NOT;
   }
 }
    #+end_src