jf@ap.co.umist.ac.uk (John Forrest) (03/10/91)
The basic scenario is this: I've recently learned, from advice Neil Rickert gave on the news, that there is an alternative to the standard address convention (ie. giving someones address or the address of a mailing list) such that you can give the name of a news group and then a list of the recipients. Something like: mailing.list:user1, user2, user3; (I don't even know if it is comma or space separated). Anyway, there is a limted version of this which looks like: a_user:; The implication being that "a_user" doesn't actually exist. This might seem pretty silly, but I have a real use for it. In our institute only a small subset of people use e-mail. If I want to send an message over e-mail to someone and a copy to someone else, there is a good probability that one of them won't use. It's still nice to use e-mail where possible though. Thus I want to comment on the mail the *people* it is going to, and if I have to I then get hard copy, and put it in the internal mail. This mechanism seemed ideal for doing this. The problem? When I tried to do it from Eudora it didn't work. It would appear that Eudora is not clever enough to realise that it is a funny mail, and it is feeding it into the SMTP envelope. Our Sendmailv5.65a+IDA does not like this syntax in the envelope - Neil Rickert has assured me it works under MH, although I haven't tried it, which presumably is a bit more intelligent in its envelope creation. The simplest thing here to fix seems to be the IDA Sendmail.mc. I've added a patch for the case I really want (ie. the :; form), which does an implied alias to "nobody" (optionally of course), and thus throws that route away. While I was doing this, it seemed to be a good idea to try to supply the more general form. However, this is much harder. Has anyone any ideas on how to do it - basically given: R$*:$+; you need to send all the parts of $2 as separate addresses. Short of spawning another sendmail and passing these addresses, I've no real ideas on how to do this! Has anyone else tried to do the same thing? John Forrest Dept of Computation UMIST PS. Thanks to all those who sent me replies on the mailq problem I posted about a week or so ago. I now think I might have been seeing an Apollo problem - ie. something locked up on one of the nodes, possibly because of the clock. The code is very much there for allowing several nodes to share a mailq file - at least in more standard Unix. I think the error I was seeing about "failure to create" was actually that where the queue daemon was operating when another node was still delivering the mail. This obviously happens periodically - the fault in this case being that one node had basically locked up, so the messages just persisted.