Early History of mh-e

[previous] [next] [table of contents] [index]

mh-e was originally written by Brian Reid in 1983 and has changed hands twice since then. Jim Larus wanted to do something similar for GNU Emacs, and ended up completely rewriting it that same year. In 1989, Stephen Gildea picked it up and is now currently improving and maintaining it.

From Brian Reid

One day in 1983 I got the flu and had to stay home from work for three days with nothing to do. I used that time to write MHE. The fundamental idea behind MHE was that it was a "puppeteer" driving the MH programs underneath it. MH had a model that the editor was supposed to run as a subprocess of the mailer, which seemed to me at the time to be the tail wagging the dog. So I turned it around and made the editor drive the MH programs. I made sure that the UCI people (who were maintaining MH at the time) took in my changes and made them stick.

Today, I still use my own version of MHE because I don't at all like the way that GNU mh-e works and I've never gotten to be good enough at hacking Emacs Lisp to make GNU mh-e do what I want. The Gosling-emacs version of MHE and the GNU Emacs version of mh-e have almost nothing in common except similar names. They work differently, have different conceptual models, and have different key bindings.

(After reading this article, I questioned Brian about his version of MHE, and received some great ideas for improving mh-e such as a dired-like method of selecting folders; and removing the prompting when sending mail, filling in the blanks in the draft buffer instead. I passed them to Stephen Gildea, the current maintainer, and he was excited about the ideas also. Perhaps one day, mh-e will again resemble MHE, although none of these ideas are manifest in Version 5.0.)

Brian Reid, June 1994

From Jim Larus

Brian Reid, while at CMU or shortly after going to Stanford, wrote a mail reading program called MHE for Gosling Emacs. It had much the same structure as mh-e (i.e., invoked MH programs), though it was simpler and the commands were slightly different. Unfortunately, I no longer have a copy so the differences are lost in the mists of time.

In '82-83, I was working at BBN and wrote a lot of mlisp code in Gosling Emacs to make it look more like Tennex Emacs. One of the packages that I picked up and improved was Reid's mail system. In '83, I went back to Berkeley. About that time, Stallman's first version of GNU Emacs came out and people started to move to it from Gosling Emacs (as I recall, the transition took a year or two). I decided to port Reid's MHE and used the mlisp to Emacs Lisp translator that came with GNU Emacs. It did a lousy job and the resulting code didn't work, so I bit the bullet and rewrote the code by hand (it was a lot smaller and simpler then, so it took only a day or two).

Soon after that, mh-e became part of the standard Emacs distribution and suggestions kept dribbling in for improvements. mh-e soon reached sufficient functionality to keep me happy, but I kept on improving it because I was a graduate student with plenty of time on my hands and it was more fun than my dissertation. In retrospect, the one thing that I regret is not writing any documentation, which seriously limited the use and appeal of the package.

In '89, I came to Wisconsin as a professor and decided not to work on mh-e. It was stable, except for minor bugs, and had enough functionality, so I let it be for a few years. Stephen Gildea of BBN began to pester me about the bugs, but I ignored them. In 1990, he went off to the X Consortium, said good bye, and said that he would now be using xmh. A few months later, he came back and said that he couldn't stand xmh and he asked if I could put a few more bug fixes into mh-e. At that point, I had no interest in fixing mh-e, so I gave the responsibility of maintenance to him and he has done a fine job since then.

Jim Larus, June 1994

From Stephen Gildea

In 1987 I went to work for Bolt Beranek and Newman, as Jim had before me. In my previous job, I had been using RMAIL, but as my folders tend to run large, I was frustrated with the speed of RMAIL. However, I stuck with it because I wanted the GNU Emacs interface. I am very familiar and comfortable with the Emacs interface (with just a few modifications of my own) and dislike having to use applications with embedded editors; they never live up to Emacs.

MH is the mail reader of choice at BBN, so I converted to it. Since I didn't want to give up using an Emacs interface, I started using mh-e. As is my wont, I started hacking on it almost immediately. I first used version 3.4m. One of the first features I added was to treat the folder buffer as a file-visiting buffer: you could lock it, save it, and be warned of unsaved changes when killing it. I also worked to bring its functionality a little closer to RMAIL. Jim Larus was very cooperative about merging in my changes, and my efforts first appeared in version 3.6, distributed with Emacs 18.52 in 1988. Next I decided mh-e was too slow and optimized it a lot. Version, 3.7, distributed with Emacs 18.56 in 1990, was noticeably faster.

When I moved to the X Consortium I became the first person there to not use xmh. (There is now one other engineer there using mh-e.) About this point I took over maintenance of mh-e from Jim and was finally able to add some features Jim hadn't accepted, such as the backward searching undo. My first release was 3.8 (Emacs 18.58) in 1992.

Now, in 1994, we see a flurry of releases, with both 4.0 and 5.0. Version 4.0 added many new features, including background folder collection and support for composing MIME messages. (Reading MIME messages remains to be done, alas.) While writing this book, Bill Wohler gave mh-e its closest examination ever, uncovering bugs and inconsistencies that required a new major version to fix, and so version 5 was released.

Stephen Gildea, June 1994

[Table of Contents] [Index] [Previous: The History of exmh] [Next: Changes to mh-e]

Last change $Date: 1996/06/06 15:09:57 $

This file is from the third edition of the book MH & xmh: Email for Users & Programmers, ISBN 1-56592-093-7, by Jerry Peek. Copyright © 1991, 1992, 1995 by O'Reilly & Associates, Inc. This file is freely-available; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation. For more information, see the file copying.htm.

Suggestions are welcome: Jerry Peek <jpeek@jpeek.com>