RE: New Standard for Whois
"Lu, Ping" <[email protected]>
| Newsgroups | gmane.ietf.whois |
|---|---|
| Message-ID | <[email protected]> |
While I would like to have XML-like capability to do a "whois-ng" query, I don't see how can you mix "whois" with "XML". Here are the reasons: 1. whois requires CRLF to be the terminator of a query. 2. XML uses <> as the delimitor and it is possible to have CRLF as part of data. 3. How can you ever escape the CRLF and not breaking the current whois implementation when you telnet into a whois server ? 4. The other way around is also difficult. How can you make sure some whois query that contains <...> is not mistaken as an XML tag ? Ping Lu Cable & Wireless USA Network Tools and Analysis Group W: +1-703-292-2359 E: [email protected] > -----Original Message----- > From: Shane Kerr [mailto:[email protected]] > Sent: Friday, January 26, 2001 9:21 AM > To: Randy Bush > Cc: [email protected] > Subject: Re: New Standard for Whois > > > On Tue, 23 Jan 2001, Randy Bush wrote: > > > > If the Whois stays on Port 43, backwards compatibility is > maintained > > > by the following approach. > > > > > > 1. Accept a connection. > > > 2. Parse the result. If XML has been received, go to step 4. > > > 3. Not XML: as per existing Whois standard, return > the result and > > > close the connection. > > > 4. XML has been received: must be the new protocol. > Maintain the > > > connection, and use the RRP. Access will probably be > > > restricted, and it may not be possible to modify objects. > > > > what if mirjam and i have an existing application running over > > whois/43 that uses xml? > > The proposal is to format the Whois *query* as XML. The only way an > existing Whois server can do this and still be RFC 954 > compliant (now a > draft standard for over 15 years!) is by sending the entire > XML document > as a single line: > > Connect to the SRI-NIC service host at TCP service port 43 > (decimal). > > Send a single "command line", ending with <CRLF> (ASCII CR and > LF). > > Receive information in response to the command line. The server > closes its connection as soon as the output is finished. > > While possible, does this ever really happen? I expect not. > > Not that this means that I think the proposal is a good idea.... > > -- > Shane > >