checksum from the file is returned via the getDatasetFiles() hessian method in an xml attribute encoded as <cheksum_type>="<checksum_value>", like:

<file MD5="7d4806ffb0e62530a2efe54724fe2a0a" ...

This is not very robust as we are not limited to md5.

I'd propose expanding this to: checksum_type="<checksum_type>" checksum_value="<checksum_value>", e.g. checksum_type="MD5" checksum_value="7d4806ffb0e62530a2efe54724fe2a0a".

Even if only MD5 is supported, it might be better to encode it in something like "checksum" instead of using the algorithm name