Bug with #206

[ArvidJB]

There seems to be a problem with the changes introduced in #206:

In [8]: with h5py.File('/tmp/foo.h5', 'w') as f:
   ...:     vf = VersionedHDF5File(f)
   ...:     with vf.stage_version('0') as sv:
   ...:         sv.create_dataset('bar', data=np.arange(10))
   ...:

In [9]: with h5py.File('/tmp/foo.h5', 'r+') as f:
   ...:     vf = VersionedHDF5File(f)
   ...:     with vf.stage_version('1') as sv:
   ...:         sv['bar'].resize((12,))
   ...:         sv['bar'][8:12] = sv['bar'][6:10]
   ...:         sv['bar'][6:8] = [0, 0]
   ...:
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
/opt/python/python-3.7/lib64/python3.7/_collections_abc.py in setdefault(self, key, default)
    853         try:
--> 854             return self[key]
    855         except KeyError:

/codemill/bessen/versioned_hdf5_env/lib64/python3.7/site-packages/versioned_hdf5/hashtable.py in __getitem__(self, key)
    103     def __getitem__(self, key):
--> 104         i = self._indices[key]
    105         shapes = self.hash_table['shape']

KeyError: b'w=\x14\x96<\xa5\xd2\xcc\xc1\xd7o/U\x0e\xc3j[*t\x8c"p\x05\xd2\xc0d\x9dt\xc1?\xdb\xe6'

During handling of the above exception, another exception occurred:

IndexError                                Traceback (most recent call last)
<ipython-input-9-2157bc8c6ae6> in <module>
      4         sv['bar'].resize((12,))
      5         sv['bar'][8:12] = sv['bar'][6:10]
----> 6         sv['bar'][6:8] = [0, 0]
      7

/opt/python/python-3.7/lib64/python3.7/contextlib.py in __exit__(self, type, value, traceback)
    117         if type is None:
    118             try:
--> 119                 next(self.gen)
    120             except StopIteration:
    121                 return False

/codemill/bessen/versioned_hdf5_env/lib64/python3.7/site-packages/versioned_hdf5/api.py in stage_version(self, version_name, prev_version, make_current, timestamp)
    201                            compression=group.compression,
    202                            compression_opts=group.compression_opts,
--> 203                            timestamp=timestamp)
    204         except:
    205             delete_version(self.f, version_name, old_current)

/codemill/bessen/versioned_hdf5_env/lib64/python3.7/site-packages/versioned_hdf5/versions.py in commit_version(version_group, datasets, make_current, chunks, compression, compression_opts, timestamp)
    121             if chunks[name] is not None:
    122                 raise NotImplementedError("Specifying chunk size with dict data")
--> 123             slices = write_dataset_chunks(f, name, data, shape=shape)
    124         elif isinstance(data, InMemorySparseDataset):
    125             write_dataset(f, name, np.empty((0,)*len(data.shape),

/codemill/bessen/versioned_hdf5_env/lib64/python3.7/site-packages/versioned_hdf5/backend.py in write_dataset_chunks(f, name, data_dict, shape)
    166                 raw_slice = Slice(idx*chunk_size, idx*chunk_size + data_s.shape[0])
    167                 data_hash = hashtable.hash(data_s)
--> 168                 raw_slice2 = hashtable.setdefault(data_hash, raw_slice)
    169                 if raw_slice2 == raw_slice:
    170                     data_to_write[raw_slice] = data_s

/opt/python/python-3.7/lib64/python3.7/_collections_abc.py in setdefault(self, key, default)
    854             return self[key]
    855         except KeyError:
--> 856             self[key] = default
    857         return default
    858

/codemill/bessen/versioned_hdf5_env/lib64/python3.7/site-packages/versioned_hdf5/hashtable.py in __setitem__(self, key, value)
    129             self.hash_table[self._indices[key]] = kv
    130         else:
--> 131             self.hash_table[self.largest_index] = kv
    132             self._indices[key] = self.largest_index
    133             self.largest_index += 1

IndexError: index 1 is out of bounds for axis 0 with size 1

Can you take a look?

[asmeurer]

Oops. This is one of those bugs that the test suite didn't catch because it only triggers when you close the file and reopen it. Fix at #209.

[ArvidJB]

Thanks @asmeurer this change seems to fix all issues we encountered.