GRZ027@DBNGMD21 (Peter Sylvester +33 1 69823973) (04/16/91)
Some recent discussions about mapping (and some examples) reminded me that I wanted to ask whether the controversal discussion between Christian Huitema and Steve Kille about "How to map an X.400 printable string blank to RFC822" has come to some end, or not? I would like to mention that the problem is important in my view. I know some scenarios (and these are known things to probably everybody, so please excuse if I bother some of you with old stuff): The main scenario is: Three gateways, one from X.400 to UUCP using the restricted character set mapping, one from X.400 to "real" RFC822 using the full character set mapping, and last and least, one of the usual RFC822-uucp "gateways" that actually do nothing. You can now easily imagine what will happen with X.400 blanks and uucp underscores and RFC822 underscores and blanks travelling through several gateways and back in different orders. I am not sure whether I fully understand all positions on this problem: RFC1148 no longer has the restricted character set mapping in it (in order to irritate the unix people). Nevertheless there is a number of RFC987 gateways that use the restricted characyer set mapping (and RFC1148 gateways can have the full to restricted character set mapping). So where are we now? Is it necessary to vote by feet (i.e. by counting the number of implementations that do it this way or the other way?) If we adopt the full character set mapping, what would it break? As long as the mail doesn't travel through uucp, nothing. Or do we prefer the restricted character set mapping, and always avoid quoted strings? There are RFC822 systems that have their ids with underscores, name_given or else, if this person sends a message to two recipients, and on one paths it happens that a restricted character set gateway is used and on the other path a full character set gateway is used, we either get a blank or (u) (most likely the best mapping would have been something to surname and givenname but that's another story). One can try and tell internet sites: Avoid underscores in the local part of your addresses. Can we also tell X.400 people to avoid blanks. Is there a solution with some heuristics? I was almost on Steve Kille side until I saw the recent message about blanks in a prmd name. Here an argument: One intention of a prmd attribute is that you can use it for incremental routing in X400. This functionality should be carried over also to the Internet environment, and therefore it must be possible to map the values to a routable entity, in this case it would be some domain name. By mapping the name to a local part /prmd=this company/ one would loose any possibility to use the data for routing (unless ...) Peter Sylvester -- EARN Office Paris