[jdev] An old pseudo standard, but is it still good?

Carlo v. Loesch CvL at mail.symlynX.com
Wed Nov 15 02:05:42 CST 2006

Maciek Niedzielski typeth:
| As far as I know, % was used in the same way in very old times, in SMTP,
| in cases when it was impossible to send email directly to
| fakeuser at ninja.com. (Don't ask me why why it was impossible, I don't
| know the story well - maybe it was internal domain, or maybe not
| compatible server, I don't know)

It comes from the days when mailers weren't always reachable by a
connect() to a certain host and port. For instance if you had to reach
a person on BITNET you would use userid%bitnetnode at gateway where gateway
was a machine capable to route both SMTP and bitnet email.

Also on UUCP there were hub nodes which were generally known, and
everyone knew how to route to them, but the leaves off the tree weren't,
especially if they were private machines standing in homes. So you
would do something like me%mycoolhost at majorknownhub.tld

Yes at a certain point they started appending top level domains so
that the addresses, even if delivered by dialup UUCP, were valid
addresses on internet aswell. I guess it took until the advent of
DSL to replace most UUCP on earth by TCP/IP. In developing countries
UUCP is still often essential, so you can't say that syntax is obsolete.

On the technical issue of user%host at msn.transport i must say I really
dislike how Jabber cannot deal with a clean transparent notation
like msn:user at host. These kludgy jid-deformations are both ugly and
hard to understand for the end user. Why should she put the hostname
into the user field and a fake hostname into the hostname field,
when adding an MSN buddy to her roster? Also icq:12345 would be much
nicer than the kludge that transports provide. Additionally, you
leave it to the server to route to the transport. Why should you have
a lot of new friendships just because you switched to a different
transport (= the one you had broke down or you no longer trust his
privacy promises).

I strongly support the idea of abandoning the user at host requirement
for jids and let jids be opaque strings to be interpreted by the
receiving entity.

--» Carlo v. Loesch --» http://symlynX.com » psyc://psyced.org/~lynX
	        xmpp:lynX at ve.symlynX.com » irc://psyced.org/#symlynX
	      the CryptoChat » https://psyced.org/LynX/?room=symlynX

More information about the JDev mailing list