The script should be written to use 'file' not just parse the extension.
Why? Just because in <.001% of the time the extension will be wrong? Not exactly an efficient use of resources.
Are you in that much of a hurry to get your logwatch reports?
I'd prefer to have a more accurate program than knowingly produce an incomplete program just because it only affects a small number of people.
At the least, they should parse the compresscmd to see if it matches 'bzip2', which defaults to producing bz2 extensions.
Thanks, Alex