gnu@sun.uucp (John Gilmore) (08/28/84)
We had the same problem as UCSD. Our current solution is to use subdomains internally (eg, I am gnu@l5.sun.uucp) and translate these to uucp-like addresses when going to the outside world (eg sun!l5!gnu). We also accept these forms, if you can get them thru other peoples' mailers to us: gnu@l5 gnu%l5 l5!gnu l5:gnu l5^gnu In general, people use "gnu@l5" internally, while outside uucp sites use "sun!l5!gnu" or just "sun!gnu" since our uucp gateway knows where we all want to get mail anyway. I'd suggest NOT generating addresses containing "%" when there is an alternative, "!". % came into use as a replacement for "@" when RFC822 decreed that you could not have two @'s in an address. As more sites start using it, the parsing becomes impossible, e.g. if you get the address ucsd!foo!bar%fly do you send it fly or foo? You can send it to fly, assuming that fly connects to foo via uucp, or you can send to foo, assuming that fly shares an Ethernet with it. Either way you'll be wrong half the time. If all you parsed was "!" (from the left), there would be no problem. The set of addresses you accept isn't half as important as the set you generate as From addresses, since that determines who can respond to you. Most mail is responses, plus people kind of expect them to work, better than a path they dreamed up looking at a net map, anyway.