When using the DumpLogSegments with the "--offsets-decoder" option (for consuming __consumer_offsets), the encoding of group metadata makes it a little difficult to identify individual fields. In particular, we use the following formatted string for group metadata:
Keys have a similar formatting. Most users are probably not going to know which field is which based only on the output, so it would be helpful to include field names. Maybe we could just output a JSON object?