From 2b5dac4d267dff5cee517aa32cf960e7826fddb0 Mon Sep 17 00:00:00 2001
From: Roland Haas <roland.haas@physics.gatech.edu>
Date: Wed, 5 Sep 2012 10:13:52 -0400
Subject: CarpetIOHDF5: make index files sparse files with full datasets

rather than datasets of extent [1,1,1]. It so happens that this patch removes
the separate shape arrays and separate dataspaces for the datasets in index
files and creates the index file datasets with the same dataspace as the
"heavy" file datasets. It retains the h5shape attribute that was originally
introduced for index files even though it is now redundant (since one could
call H5Sget_simple_extent_dims on the datasets in the index files).
---
 Carpet/CarpetIOHDF5/src/Output.cc | 18 ++++--------------
 1 file changed, 4 insertions(+), 14 deletions(-)

diff --git a/Carpet/CarpetIOHDF5/src/Output.cc b/Carpet/CarpetIOHDF5/src/Output.cc
index 1df156c16..44f9df478 100644
--- a/Carpet/CarpetIOHDF5/src/Output.cc
+++ b/Carpet/CarpetIOHDF5/src/Output.cc
@@ -354,13 +354,11 @@ int WriteVarChunkedSequential (const cGH* const cctkGH,
 
       // Get the shape of the HDF5 dataset (in Fortran index order)
       hsize_t shape[dim];
-      hsize_t index_shape[dim];
       hsize_t num_elems = 1;
       for (int d = 0; d < group.dim; ++d) {
         assert (group.dim-1-d>=0 and group.dim-1-d<dim);
         shape[group.dim-1-d] = (bbox.shape() / bbox.stride())[d];
         num_elems *= shape[group.dim-1-d];
-        index_shape[group.dim-1-d]  = 1;
       }
 
       // Don't create zero-sized components
@@ -443,7 +441,7 @@ int WriteVarChunkedSequential (const cGH* const cctkGH,
           }
 
           // Write the component as an individual dataset
-          hid_t plist, dataspace, dataset, index_dataspace, index_dataset;
+          hid_t plist, dataspace, dataset, index_dataset;
           HDF5_ERROR (plist = H5Pcreate (H5P_DATASET_CREATE));
           // enable compression if requested
           const int compression_lvl = request->compression_level >= 0 ?
@@ -463,10 +461,8 @@ int WriteVarChunkedSequential (const cGH* const cctkGH,
                                            filedatatype, dataspace, plist));
 
           if (indexfile != -1) {
-            HDF5_ERROR (index_dataspace = H5Screate_simple (group.dim,
-                                                            index_shape, NULL));
             HDF5_ERROR (index_dataset = H5Dcreate (indexfile, datasetname.str().c_str(),
-                                                   filedatatype, index_dataspace, H5P_DEFAULT));
+                                                   filedatatype, dataspace, H5P_DEFAULT));
           }
 
           io_bytes +=
@@ -481,7 +477,6 @@ int WriteVarChunkedSequential (const cGH* const cctkGH,
           HDF5_ERROR (H5Dclose (dataset));
 
           if (indexfile != -1) {
-            HDF5_ERROR (H5Sclose (index_dataspace));
             error_count += AddAttributes (cctkGH, fullname, group.dim,refinementlevel,
                                           request, bbox, index_dataset, true);
             HDF5_ERROR (H5Dclose (index_dataset));
@@ -620,17 +615,15 @@ int WriteVarChunkedParallel (const cGH* const cctkGH,
 
       // Get the shape of the HDF5 dataset (in Fortran index order)
       hsize_t shape[dim];
-      hsize_t index_shape[dim];
       hssize_t origin[dim];
       for (int d = 0; d < group.dim; ++d) {
         assert (group.dim-1-d>=0 and group.dim-1-d<dim);
         origin[group.dim-1-d] = (bbox.lower() / bbox.stride())[d];
         shape[group.dim-1-d]  = (bbox.shape() / bbox.stride())[d];
-        index_shape[group.dim-1-d]  = 1;
       }
 
       // Write the component as an individual dataset
-      hid_t plist, dataspace, dataset, index_dataspace, index_dataset;
+      hid_t plist, dataspace, dataset, index_dataset;
       HDF5_ERROR (plist = H5Pcreate (H5P_DATASET_CREATE));
       // enable compression if requested
       const int compression_lvl = request->compression_level >= 0 ?
@@ -650,10 +643,8 @@ int WriteVarChunkedParallel (const cGH* const cctkGH,
                                        filedatatype, dataspace, plist));
 
       if (indexfile != -1) {
-        HDF5_ERROR (index_dataspace = H5Screate_simple (group.dim,
-                                                        index_shape, NULL));
         HDF5_ERROR (index_dataset = H5Dcreate (indexfile, datasetname.str().c_str(),
-                                               filedatatype, index_dataspace, H5P_DEFAULT));
+                                               filedatatype, dataspace, H5P_DEFAULT));
       }
 
       io_bytes +=
@@ -667,7 +658,6 @@ int WriteVarChunkedParallel (const cGH* const cctkGH,
       HDF5_ERROR (H5Dclose (dataset));
 
       if (indexfile != -1) {
-        HDF5_ERROR (H5Sclose (index_dataspace));
         error_count += AddAttributes (cctkGH, fullname, group.dim,refinementlevel,
                                       request, bbox, index_dataset, true);
         HDF5_ERROR (H5Dclose (index_dataset));
-- 
cgit v1.2.3