I use PyTables and often make use of CArray and EArray classes for their support of compression.
I'm working on a project that may need compatibility outside of Python (in Java, most likely), however. Does that mean I can't use CArray and EArray? Or is there a library for handling these?
HDF5 supports both compressed (what PyTables calls CArrays) and extendible (what PyTables calls EArrays) datasets. These are general features available in all implementations of HDF5.
However, there is one thing to be aware of - it depends on the compression algorithm that PyTables uses. h5py, another HDF5 library for python, has the option to use LZF compression, which is not supported by all HDF5 installations. If it's using gzip, then the arrays should be readable everywhere.
You can easily check the portability of existing files with the free HDFView tool, which uses Java.
