1 \chapter{Introduction}

2 \label{chap:introduction}

3

4 \section{History of electronic mail}

5 %TODO: have a quote from Bell Labs about email here

6 %FIXME: add references to text

7

8 Electronic mail (short: \name{email}) is a basic concept in \unix. On \unix\ machines, a lot of information is distributed by \name{system mail}, which is email sent by the operating system. Beside that, email is the common communication system between humans working on computers.

9

10 The \unix\ operating system supports email since 1979 through the \name{mail user agent} (\NAME{MUA}) \path{/bin/mail}. For transporting mail in between two systems, the \NAME{UUCP} protocol (for ``\unix\ to \unix\ copy'') was invented.

11 %FIXME: what about `uuxqt' and `rmail'

12

13 As generally known, the early development of \unix\ was not only made in the \name{Bell Labratories} of \NAME{AT\&T}. But also the \name{Univerity of California at Berkeley} worked on their version of a \unix\ operating system, called \NAME{UCB} \unix, or \name{Berkeley} \unix.

14

15 This lead to a second \NAME{MUA} from Berkeley: \name{Mail} (with a capital `M'). Also, no \NAME{UUCP} network was set up at Berkeley but an own creation called \name{BerkNet} was used.

16 %FIXME: why? license issues? closed source?

17 %TODO: what about `mailx'?

18

19 Further more there was a third network type: the \NAME{ARPAnet}, based on the \name{transmission control protocol} (\NAME{TCP}).

20 %FIXME: where did it came from?

21

22 Each of the three network types could transfer email between different machines. The file transfer itself was made using \NAME{FTP}, but the higher layered logic of the transfer was different in each. For example was the addressing schema not the same: \NAME{UUCP} used a flat-style schema, while \NAME{ARPAnet} was hierachical.

23

24 Mail transport from one machine using one kind of network to a second machine using another kind was a problem. This showed up at Berkeley where some departments of the university had switched to \NAME{ARPAnet}, and some to \NAME{UUCP}, while the rest was \name{BerkNet}.

25

26 It was around 1982, when Eric Allman, then a student at Berkeley, wrote \name{delivermail}. Its purpose was to transform email from one network to another. \name{delivermail} like its successor, the more flexible \sendmail, intermediated between the different networks. They were able to transform email messages from any network to any other.

27

28 Todays email structure is basicly the same as then. The major difference is the uniformity of the underlying network, which is nearly always the \NAME{ARPAnet}-based \name{Internet}. Most other differences are organized as extensions to the old model of electronic mail.

29

30 More information about electronic mail and its history can be found at:

31 %FIXME: add books and websites here

32

33

34

35 \section{Transporting mail}

36 %TODO: include definitions from others here (cites)

37 %TODO: when was the term ``mail transfer agent'' established?

38 The basic job of a \name{mail transfer agent} (or \name{mail transport agent}, short \NAME{MTA}) is to transfer/transport \name{electronic mail} (short \name{email}) from one host to another.

39

40

41

42

43 \section{\sendmail}

44 \label{sec:sendmail}

45 \name{sendmail} is the de facto standard for \name{mail transfer agents}.

46

47 % FIXME: is that true?

48 It was the first \NAME{MTA} and had no real alternative for a long time.

49

50 All other existing substitutes, which are mainly \name{postfix}, \name{exim}, \name{qmail} and the here regarded \name{masqmail}, mimic \name{sendmail}'s behavior. Especially, they all create a symbolic link named ``sendmail'' pointing to their own executable. This is because a lot of programs assume there is an executable called ``sendmail'' on every computer system.

51

52 Besides being the ``standard'', \name{sendmail} probably is the most scalable and powerful solution for transfering emails and definatly the most flexible one.

53

54

55

56 \section{(a look at Windows)}

57