The way a machine name is interpreted, i.e. at which site your message
will finally wind up, and how to combine this name with the recipient's
user name greatly depends on the network you are on. In the original UUCP
environment, the prevalent form was path!host!user,
where path described a sequence of hosts the message had to travel
before reaching the destination host. This construct is called the
bang path notation.
Internet sites adhere to the RFC 822 standard, which uses a notation of user@host.domain, where host.domain is the host's fully qualified domain name. The middle thing is called an ``at'' sign. Because this notation does not involve a route to the destination host, but gives the (unique) hostname instead, this is called an absoluteaddress. Today, many UUCP-based networks have adopted RFC 822, and will understand this type of address.
Now, these two types of addressing don't mix too well. Assume an address of hostA!user@hostB. It is not clear whether the `@' sign takes precedence over the path, or vice versa: Do we have to send the hostB, which mails it to hostA!user, or should it be sent to message to hostA, which fowards it to user@hostB?
Addresses that mix different types of address operators are called hybrid addresses. Most notorious is the above example. It is usually resolved by giving the path precedence over the `@' sign. In the above example, this means sending the message to hostA first.
However, there is a way to specify routes in RFC 822-conformant ways: <@hostA,@hostB:user@hostC> denotes the address of user on hostC, where hostC is to be reached through hostA and hostB (in that order). This type of address is freqeuently called a route-addr address.
Then, there is the `%' address operator: user%hostB@hostA will first be sent to hostA, which expands the rightmost (in this case, only) percent sign to an `@' sign. The address is now user@hostB, and the mailer will happily forward your message to hostB which delivers it to user. This type of address is sometimes referred to as ``Ye Olde ARPANET Kludge''. It was used on the ARPANET to reach hosts who were not registered officially with the Network Information Center, so that messages had to be sent via a mail relay that was known to handle mail for these ``hidden'' sites.
Other networks have still different means of addressing. DECnet-based networks,
for example, use two colons as an address separator, yielding an address
of host::user.
Lastly, the X.400 standard uses an entirely
different scheme, by describing a recipient's by a set of attribute-value
pairs, like country and organization.
On FidoNet, each user is identified by a code consisting of four numbers,
denoting country, node, point, and user.
There are some implications to using these different types of addressing which will be described throughout the following sections. In a RFC 822 environment, however, you will rarely use anything else than absolute addresses like user@host.domain.