* Ability to register types as "YAML", so that they can be read from the HDF5 ``attrs`` (or "meta") as native types via YAML. This works both ways (i.e. read and write) as long as a constructor/representer is defined for the object in YAML.
2.4.0
======
Added ----- * Ability to open ``HDF5Object`` instances' files as writeable, so as to append extra data to them.
2.3.9
======
Try again to get this on PyPI... grrr :-(
2.3.7
======
A whole bunch of minor updates just to get the auto-deployment working...
2.3.1
======
First version on PyPI!
2.0.0
====== Changed ------- * Removed ``.load`` and ``.load_all`` methods from ``HDF5Object`` so that the interface between the object and its sub-groups are more identical. Note that *everything* now caches when accessed, but can be explicitly removed from the cache at will.