![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to p1 CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [CSRG BSD Unix] / 43BSDReno / share / doc / ucs / mmdf|
Return to p1 CVS log | Up to [CSRG BSD Unix] / 43BSDReno / share / doc / ucs / mmdf |
1.1 root 1: .SH
2: Introduction and History
3: .PP
4: The Multi-channel Memo Distribution Facility,
5: commonly called MMDF,
6: is a suite of software that has seen a great deal of work since it was
7: originally released in 1980.
8: The original code was designed and implemented by Dave Crocker
9: working under Professor David Farber at the University
10: of Delaware (UDEL).
11: The MMDF system was then chosen to form the initial backbone
12: software for the CSNET project and has been in use for several years
13: by elements of the U.S. Army.
14: The software has seen a great deal of change in the process.
15: The original code is commonly referred to as MMDFI or MMDF Version 1.
16: A number of minor additions and changes were made while fielding MMDFI
17: as the result of collaboration between UDEL and BRL and some other
18: sites.
19: After the original code was fielded in CSNET, Dave Crocker began the
20: development of a upgraded version of the MMDF system which was
21: designed to work in the new Internet
22: domain naming system
23: and was to incorporate numerous design changes suggested by
24: experience with MMDFI.
25: Dave Crocker
26: left the CSNET project before completing this work, approximately
27: two weeks before the TCP/IP switchover of the ARPANET, 1 January 1983.
28: At this time, BRL was a solid MMDF site.
29: We were reluctant to try to retrofit
30: the existing version of MMDFI
31: to handle the new mail protocols that also took effect on
32: 1 January, so Doug Kingston of
33: BRL undertook the task of finishing the work
34: needed to make MMDFII operational.
35: A production version of MMDFII was installed at BRL during
36: the third week of January 1983, and served
37: as BRL's mail system on three hosts,
38: but there was no stable version of the MMDFII code until June 1983.
39: The first few months of MMDFII
40: were quite rough and it needed a great deal of ``tender loving care''.
41: .PP
42: For reasons that will be clear in a moment, this stable version of June 1983
43: is now referred to as the MMDFII-pre-England version.
44: Around June, a copy of this stable version was delivered to Steve Kille
45: of University College London (UCL) and to Brendan Reilly
46: of UDEL, who
47: had taken over Dave Crocker's work on MMDF at UDEL.
48: Steve Kille made a number of major changes to the handling of domains,
49: address parsing, and handling of the alias files.
50: Steve also added support for NIFTP, a European file transfer protocol
51: used for sending mail in a batch environment.
52: At the same time that
53: Steve was making his enhancements, Doug Kingston continued
54: to develop BRL's copy of MMDFII to make it an even more solid
55: mail system. BRL's changes were not as major as Steve's
56: but covered a great deal of code and fixed several major outstanding bugs.
57: This dual development led to two variants of MMDFII that each
58: needed the other's improvements.
59: In late September of 1983
60: Brendan Reilly and Doug Kingston spent a week in England with Steve
61: to merge the variants and to discuss future changes and directions
62: for MMDF. The result of this meeting was a merged version of
63: MMDFII which I will call MMDFII-post-England.
64: Just prior to this trip, the CSNET Information Center (CIC)
65: received a copy of the pre-England MMDF.
66: Their later changes were based on this pre-England version
67: which made merging of their changes into the
68: post-England version somewhat difficult.
69: .PP
70: After the England meeting, Brendan Reilly of UDEL took the role
71: of coordinator of the subsequent changes to MMDF. Copies of the
72: MMDF-post-England were made simultaneously available to BRL, UCL, and UDEL.
73: Since then many minor changes have been made by all
74: four sites;
75: in essentially all cases these changes have been bug fixes or changes
76: to make MMDF a more stable and robust system.
77: .PP
78: Since then, Doug Kingston at BRL has made changes
79: to the local delivery mechanism, rewriting much of the original
80: code, and the central delivery program has been upgraded to
81: take advantage of large-address-space machines, when possible, to
82: keep retry histories for messages on a host-by-host basis.
83: Bernie Cosell at the CIC has undertaken to speed up MMDF
84: execution by providing a facility for compiling in some of the information
85: normally included in the ASCII text-based version. Steve Kille
86: an alternative to the ASCII text based version. Steve Kille
87: has continued to refine the address handling and the British
88: ``backwards'' domain code.
89: .FN
90: The British do domains backwards. For example, if in the
91: US (Internet) we write ``[email protected]'' known
92: as ``little endian'' order, the British
93: (SERC Net)
94: write ``[email protected]'' or ``big endian'' order.
95: Put another way, ``big endians'' put the largest, most general,
96: or most significant element of the domain first. ``Little endians''
97: use the other order, with the most significant part last.
98: [See
99: .I
100: Gulliver's Travels
101: .R
102: by Joanthan Swift. The "big endian" vs. "little endian" controversy
103: was a
104: .I
105: causus belli
106: .R
107: in Lilliput.]
108: .FE
109: Brendan Reilly has made changes to the package to allow it to
110: run on the Altos system and has fixed numerous bugs in
111: the PhoneNet code.
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.