Re: Suggested extension

[Ned Freed <[e-mail]>]

Just because this particular extension doesn't offer much exposure (i.e.
the password doesn't protect anything that important) is no excuse for
deploying another clear-text-password scheme on the Internet. Such schemes
should be carefully limited to the essential cases, since fixing them
later is quite costly.

An alternative proposal: Instead of sending the password, send a hash value.
Specifically, take the password-in-clear plus the entire text of the query
and run it through a good hash function like MD5. Take the output of this
function and send that along with the query. This eliminates exposure of the
clear text password and limits a network sniffer to repeating old queries.
(It presumably already saw the answers to these queries.) If you want a
bit more security, put a time value in there and hash it too -- this can be
used to limit queries to a small time regime.

I do agree that this should only get done in the next revision of the document.
But if it gets done at all I think a hash function is the right way to go
about it.

				Ned