keith@tarantula.spider.co.UK (Keith Mitchell) (11/01/88)
From the correspondence on this subject, people appear to prefer the average Rlogin implementation over the average Telnet one because: 1) It passes terminal type 2) It bypasses the need for username/password to be entered 3) No newline etc processing is applied to data transferred 4) It has better out-of-band data handling 5) It has better flow-control handling 6) It has window-size negotiation A full (as opposed to average) Telnet implementation can provide (1), (3), (4), and (6), and the Telnet standard is sufficently general that extensions to cover (2) and (5) are certainly possible. I still feel Telnet is the protocol to go with - networking is about connecting heterogenous systems, not just Unix ones, and achieving the goal of universal connectivity/ interoperability is never going to be accomplished with non-documented standards. The problem is that although Telnet defines a general functionality, most implementations only provide a subset of this. Rlogin implements a different subset of this functionality (It has no definition distinct from the implementation). What is really wanted is an implementation which provides a more general union of both these subsets. I think the approach implementors should take therefore, is to provide both Rlogin and Telnet, but concentrate on extending Telnet so it can meet all peoples' needs. We need both in the meantime, not least because the whole situation is very much "horses for courses". (The Telnet vs Rlogin arguments strike me as very similar to those for NFS vs RFS.) The discussion on this subject has been very useful, and I will make sure it is input to the next phase of SpiderPort development. Keith Mitchell Spider Systems Ltd. Spider Systems Inc. 65 Bonnington Road 12 New England Executive Park Edinburgh, Scotland Burlington, MA 01803 +44 31-554 9424 +1 (617) 270-3510 keith@spider.co.uk keith%spider.co.uk@uunet.uu.net keith@uk.co.spider ...!uunet!ukc!spider!keith
BILLW@MATHOM.CISCO.COM (William Westfield) (11/06/88)
No one else has mentioned it, but it is worth pointing out that RLOGIN can be made more efficient than TELNET, since a TELNET implementation has to look at the data stream looking for IACs, whereas RLOGIN (or SUPDUP) doesn't. This is not an issue one most unix systems, where scanning the data stream is not the limiting factor, but it can make a big difference in a terminal server, where you have (for example) a poor 68020 trying to deal with 96 terminal lines. Bill Westfield cisco Systems. -------