We already have Snappy stream decompression, but we still need to work on compression. See #224 and #710 for context
