Description
Description
h5 has an issue reading complex datatype written in Julia.
Steps to Reproduce
- create HDF5 archive in Julia
using HDF5
h5open("archive.h5", "w") do file
g = create_group(file, "group")
g["dataset_float"] = rand(5,5)
g["dataset_complex"] = rand(5,5) + rand(5,5)im
end
- load using TRIQS/h5
from h5 import HDFArchive
with HDFArchive('archive.h5', 'r') as h5:
dataset_float = h5['group']['dataset_float']
dataset_complex = h5['group']['dataset_complex']
- h5py instead works
import h5py
hf = h5py.File('archive.h5', 'r')
dataset_float = np.array(hf['group']['dataset_float'])
dataset_complex = np.array(hf['group']['dataset_complex'])
Expected behavior: [What you expect to happen]
Actual behavior:
RuntimeError: .. Error occurred at Thu Jan 12 09:18:06 2023
.. Error .. calling C++ overload
.. h5_read_bare(group g, std::string name) -> PyObject *
.. in implementation of function _h5py.h5_read
.. C++ error was :
HDF5/Python Internal Error : can not find the numpy type from the HDF5 type
Versions
h5 version 1.1.x and unstable