vanden@brubeck.SRC.Honeywell.COM (Kurt Vandenberg) (03/16/89)
The bsd 4.3 and Sun OS 4.0 manuals refer to reliable datagram sockets but say
they are not implemented yet. Does anyone know of an implementation
(available or in the works) for reliable datagram sockets?
Kurt Vandenberg (612) 782-7341 Honeywell
UUCP : {ihnp4,philabs,umn-cs,mmm}!srcsip!vanden Systems and Research Center
INET : srcsip!vanden@umn-cs.cs.umn.edu 3660 Technology Drive
ARPA : vanden@src.honeywell.com Minneapolis, MN 55418postel@VENERA.ISI.EDU (03/17/89)
Kurt Vandenberg: Please note, "reliable datagram" is an oxymoron (like Military Intelligence). --jon.
PADLIPSKY@A.ISI.EDU (Michael Padlipsky) (03/18/89)
One might, however, achieve oxydullnormalcy by using the checksums which are, after all, available in UDP's spec. Granted, not _truly_ reliable since lacking retransmission discipline; but possibly what's meant/wanted in context--provided, of course, that anybody implements it.... (There's also the quibble that if you define by Form rather than by Content, TCP/IP "is" a reliable datagram protocol, since it/they satisfy the formal criterion for datagrammaticality of containing source and destination addresses in each transmission unit, but I'm not really in the mood for that one, so let it go.) cheers, map -------