XEmacs is Dead. Long Live XEmacs!

Steve Yegge, you are my hero!

:)

As soon as emacs gets proportional anti-aliased fonts, I'll switch. I can't go back to scratchy fixed width fonts.

installations from the debian popularity contest:

emacs21 30982
emacs22 18491
xemacs21 10795

sure, debian’s a limited sample, but damn people are still using xemacs!

shin_dig: I can’t tell why someone would want to run a text editor (as opposed to an word processor) in a proportional font — especially for programming. as for me, you can take my clean, lean, pleasantly monospaced Terminus font away from my cold, dead, RSI-damaged hands.

I've been an XEmacs user for years now, mostly out of habit more than anything else. You may have just convinced me to give GNU Emacs another go. Hopefully my options file will play nicely with it! : )

Great article, as always.

Which leads me to the question: I am not an Emacs user and I always see people like you talking about how cool the self-hosting environment is.

I also agree that IDEs like Eclipse, NetBeans, etc get in the way and I am never motivated to write a whole plugin to them.

I confess, I am a TextMate/Mac user. It is light-weight, bundles are easy to write, its reasonably customizable. More important of all: it has a Mac-alike GUI which is the utmost thing I can't stand in Emacs.

What I mean is, it made sense 20 years ago to have a terminal-non-GUI environment that is in itself almost an operating system.

But considering that we now live in a GUI world, be it Macs, Windows and even Linux with Gnome/KDE in spite of its 'terminal' heritage.

I don't need and I really don't want to have a text-based web browser within my editor. What's the point? I can't really see its design in a text-based web browser.

I really want drag & drop between multiple applications. It is a reality today that we multi-task between multiple applications that have to have integration between each other. It's just the way it is now.

So, what is really the point on Emacs? Don't get me wrong. I am really willing to try to learn it. But I never found a single article out there with a good enough reason to do so. The only thing that every article brags about is 'self-hosting', so please elaborate on how self-hosting environments can actually improve my productivity in comparison with other modern IDEs.

Best Regards

Not to belittle Jamie Zawinski at all, but wasn't he the Unix guy on the Netscape Navigator v.1 team rather than its overall author? That's the impression I get from for example nscp dorm and his Wikipedia entry.

While I totally agree with the main points of your article, I am surprised to hear that you consider XEmacs unstable. I do run it for weeks at the time without experiencing crashes. Currently using 21.4.

You already convinced me to switch away from XEmacs a few essays ago, but I haven't gotten around to it yet. But I will say, I've never had a problem with XEmacs stability. I won't say it never crashes, but it's in the single digits per year. It's more likely that Gnome will hang or something, and sometimes even then I can save XEmacs by switching to another TTY and restarting the window manager.

As for a Firefox/Emacs hybrid, what I really want is to embed an Emacs window in every textarea in Firefox. The "It's All Text!" plugin is close (with my editor set to gnuclient), but it's still just a bit too inconvenient to click the "edit" button rather than just have an Emacs buffer be right there.

It's often more fun to be captain of a small boat than a deckhand on a big ship.

If the XEmacs guys move over to GNUEmacs, will they get the same status they had under XE? Will Jamie Zawinski be co-consul with RMS? Or will he be punted somewhere lower down the pecking order, working on more peripheral stuff?

I suspect a lot of open-source projects face a similar challenge, which is why you get so many linux distros, KDE and Gnome, and about 5.6x10^7 MVC frameworks.

Firefox with text editing capabilities? You mean, a text editor built upon the mozilla platform, with extensions and stuff? Something like OpenKomodo?

Or did you really mean integrating the capabilities in Firefox itself?

AkitaOnRails:

TextMate isn't 1e-10 times as customizable as emacs. It doesn't let you dig as deep. And the results show.

A text editor should be like a text field - it doesn't need menus, it doesn't need modern GUI buttons and dialogues and stuff. Just room for text. My emacs windows are borderless, menuless, toolbarless squares of TEXT (well, on the mac they have a shadow and a menu...). And in that, I have all I need. Emacs has a UI better suited for text editing than firefox or the OSX default UI.

Steve's point is that we DO need a REAL GUI browser in emacs.

You DO have drag and drop between emacs and the outside world.

Does this answer your points?

Just trying to help, hope you find the way that's right for you,

-- Aur

@sonoflilit well, kind of.

As I said, I am not an Emacs user and that's the point: learning-curve.

I don't dispute that Emacs is more powerful than any other editor out there. I just used Textmate as an example of an editor that is nowhere near Emacs in terms of customizability, but nonetheless it gets my job done.

Java users would say the same about Eclipse.

So, what is the 'real' benefits of a self-hosting environment like Emacs? I can understand that a Lisp user love it. But what about non-Lisp lovers?

(by the way, sorry for the off-topic).

Your title stirs a tempest in a teapot. You should gone for "Emacs is Dead. Long Live Emacs!" When you're playing with fire, might as well use a flamethower. ;-)

> It won't be long, I think, before the best Java development environment on the planet is written in JavaScript.

You always amaze me with the scope of your predictions. I just hope there's no petty license squabbles along the way.

Perhaps we need to embed a self-hosting env into FF, one that doesn't require restarts. Why can't we just tack on Guile or extend JavaScript to do that? Oh, its just a "browser", not a rich client.

(GNU Emacs user 14 years and counting)

@sonoflilit: did you know that, apparently, OS X users pay actual money for the privilege of full-screen text editing? If only they knew how nice is a big black Emacs window maximized in my stumpwm…

Forgive me for this, but I have to do it:
VIM FTW!!!

Sorry :)

Is this just a sneaky way of preparing to drive us all to your NBE?

Is this just a sneaky way of preparing us to move to your NBE?

Heh. I would love some kind of "buffer like" thing with a cairo canvas, on which I could draw with elisp. -- So, where do we go to buy some free megaseconds?

There is also sxemacs project, that had forked XEmacs and try to extend it with many features, like multi-threading and so on.
GNU Emacs gain more success since 21st release, that had many features, not existing in XEmacs, and also has less troubles with multi-language environments (XEmacs all times had problems with non-latin fonts, etc - i had used XEmacs at 1997-2001)
Currently i look for a discussions about adding many IDE capabilities to GNU Emacs, and i like this approach.

It's past the time to move out of plain text and IDE's into mainstream Literate Programming.

Java should learn performance from the emacs byte-compiler. Emacs starts up a lot faster than any similarly-sized Java app; yet they are both bytecode systems.

It's also time to start mainstreaming voice input to complement the keyboard and mouse.

Bytecode systems are increasingly obsolete. It wouldn't be terribly hard to generate os-portable x86 machine code for OSX, Solaris, Linux or Windows. Only 2 kinds; 32-bit and 64-bit. Mediocre machine code will outperform bytecode.

A good, solid 90% solution would be someone with strong emacs-fu (I am
but tiny grasshopper) figuring out how to embed Gecko (WebKit would be
fine, too) into Emacs.

Gecko buffers in my buffer list would totally own. I could have one
integrated command line/address bar/search bar/find bar instead of
the several different command-line wannabees that Firefox has. And having everything in my buffer-list instead of tabs would be a towering victory.

Sure, it wouldn't have the same chrome control as XUL, the browser bits would require completely different extension techniques than the rest of emacs, and you wouldn't have the full multilingual support you're talking about, but, at least for my purposes, it would definitely quality as a 90% solution.

> Port to XUL, maybe?

Are you familiar with conkeror? It's an emacs-inspired web browser built on top of XUL. While it currently only aspires to be a web browser, it's picking up an awful lot of the emacs nature, and I suspect that if someone wanted to, it could easily turn into a real emacs, but with a superior rendering engine and Javascript as the extension language.

Also, rumor has it that Microsoft is working on their own Bizarro-world Emacs. Maybe that'll be the solution!!

From about '79, i've been using flavors of emacs, starting with FINE (Fine Is Not Emacs) and it's bigger brother COURSE. The original EMacs used ITS TECO as it's extension language. As a TECO user, it was natural to jump into EMacs, and still be able to write powerful one-off editing scripts. I recall doing a matrix transpose in TECO off the top of my head. It was five minutes to write it, and ten minutes to execute (on a .2 MIPS pdp-10).

In the '80s, my variant of choice was Jove - i had binaries that ran on the Mac, DOS and Unix, the three places i was in most. DOS died, and the Linux version became xjove, and my fingers didn't know how to use it. The original version died with glibc2. Meanwhile, in the late 80's, the 386/25 was the first machine that started up Emacs fast enough (4 to 10 MIPS, depending on how you counted it.) And then GNU Emacs kept changing the rules for how to set ^h to 'back-hacking-tabs'. Really, i want to build on what i know, not have to relearn everything i know. How am i supposed to make progress?

I've never gotten very far into Lisp and variants, though i've been goofing with Guile of late. Far from using a Turing complete language to edit my files, i'm now stranded with keyboard macros, and i tell them to run a million times, stop on ringing the bell. My Emacs experience has become so limited, i've started using Vim. And liking it. Yet, my peers know even less about editing, and want me to make the big sweeping changes, because all they know is copy and paste.

I really don't want Emacs to replace Eclipse as my operating system. Eclipse would be a good IDE, if it was slimmer. 10000x slimmer. Turbo C 2.0 worked on a 640k DOS machine. Think C was awesome on a 1 MIPS 68K Mac with 2 MB RAM. Awesome. Now i can't bring up the editor without 2 GB RAM and a cup of coffee. I don't even drink coffee. Maybe i should get a new 1 MIPS 68K Mac.

TECO is a line noise language. But it has very little indirection, so it can be learned quickly. It's a stack language, so could be thought of as Forth, but very much special purposed. It doesn't have the verbosity of Scheme, Lisp or Java. So the commands can be interpreted as is, and very fast. Who needs byte code? There was quite a bit to be said for it. I'm not suggesting that we switch back.

However, i'm convinced that the quality of Lisp code out there has more to do with the barrier to entry than anything inherent in the language. The barrier to entry is high, so the really bad drivers stay home. Of course the code is better. And Gabriel dispairs that really good looking Lisp code runs really slow. You can't tell by looking. He is correct that C programmers are unable to make this kind of mistake. And if you're not supposed to use call/cc (Scheme) because the result is unpredictable in time, space and, well, what is it going to do, then you have a language that is recursive, like C is, or at least almost as good. What of it? Why not use C, which can be compiled to the bare metal, and by the way, is still the gold standard for speed? Garbage collection? That may be all you get.

I couldn't tell you which Emacs i have on my Linux box. Did i say "sudo apt-get install emacs", or did i say "sudo apt-get install xemacs". But apparently, when i needed an editor at work, it was a Windows Emacs package that came up first in the search. It wasn't because Emacs is more reliable or has fresh minty taste. If that's why you think Emacs is becoming more popular than xemacs, you've got some more karma to work out.

There's a library called saveconf.el distributed with Xemacs that saves and retrieve buffer states. It reminds me of my beloved save-tabs-on-exit in Firefox 3b5. If you added a timer to save-context it would probably go a long way to save and recovery on crash.

I looked at desktop.el, which looks like it saves a lot more than saveconf.el. But it seems to me that a lot of what desktop.el saves is stuff that you tend to set through your .emacs anyway.

saveconf.el

Thanks for all the Emacs articles, by the way. You convinced me to switch about a month ago and it continues to be a mind-blowing experience.

> As soon as emacs gets proportional anti-aliased fonts, I'll switch. I can't go back to scratchy fixed width fonts.

It does. It's not in the last released version (emacs22), but it's in what will become emacs23, and your distribution might have packaged a prerelease version.

Screenshots:

http://peadrop.com/blog/2007/01/06/pretty-emacs/
http://times.usefulinc.com/2005/12/02-emacs-xft

I was a die-hard xemacs user from 1996 until early 2008. I was already a doubting thomas, but Aquamacs put the final stake in it (of course, RMS doesn't like Aquamacs either, but that's another story).

To the Textmate/Mac user: check out Aquamacs at aquamacs.org -- it's an OS X / Aqua port of Emacs. Very nice.

Does emacs have a Bugzilla or Trac system going yet? I remember some old argument about that hampering its development.

GNU Emacs now has debian-style bug reporting system

You found it! The way to embrace both Javascript and Emacs. You are my hero.

And you thought I'd given up on controversial blogs.
you're still mad about my OS X blog.
It's a blog I've been working on for years

A blog is a sequence of posts. Please stop using "blog" to mean "blog post". Thank you.

Other occurrences on the page that indicate you know this already:
A dude with arguably the best cat-picture blog ever created.
Keep up the cat blog.
Stevey's Personal Blog

Maybe stevey's not using the normal word "blog" (hereafter "blog¹") at all. I think he's nominalizing the verb "to blog²" to create a new noun "blog³", like this:

I borked this yesterday -> I made a bork yesterday
I grokked something today -> I had a grok today
I blogged² about OSX -> I worked on a blog³ about OSX.

That there was already a noun "blog¹" with a different meaning is irrelevant to this proccess; what matters is that the verb "to blog²" means "to create a blog¹ post", not "to create a blog¹". Thus, blog³ = blog¹ post.

Since blog¹ is actually the source of "to blog²", blog³ is a case of re-nominalization of a verbalized noun, with a shift in semantics along the way!

I'm holding out for the combined self hosting emacs, firefox, wow program.

All this time I thought XEmacs was just Emacs with a GUI front end. Thanks for clearing that up for me!

I honestly don't think that you could truly make emacs into a multi-language platform without destroying some of the 'self-hosting' properties of it. You would need a system that could intermingle function calls and variables from various languages. This strikes me as a rather convoluted task as the languages that would be good candidates, (Python, Ruby, basically dynamic languages) tend to seem to have incompatible features. I try to imagine integrating Ruby's object system into e-lisp or trying to call a macro from Python, and it seems like that for the languages to interoperate, you'd have to loose the a lot of the meta-programming techiniques from all languages which make them so popular in the first place, or, at least restrict them so they don't carry over between languages.

Aren't you using a mac now? The you might like http://aquamacs.org/

Think GNU Emacs but it behaves like a Mac application. I used it since switching and haven't been disappointed thus far.

The McCain/Obama/Clinton reference in "At this point it's becoming painful to watch. GNU Emacs is getting all the superdelegates. That warmonger VIM is sitting back and laughing at us. But XEmacs just won't quit!" was badass. You're the man, Stevie. :)

SonOfLilit said “TextMate isn't 1e-10 times as customizable as emacs.”

You're speaking from ignorance here. The declarative nature the majority of its customizations means that most TextMate bundles progress at a much faster rate than emacs modes. Its scope system make them in many ways more flexible and powerful.

The addition of tm_dialog, and built-in WebKit, mean that TextMate can already do some of the things Steve is talking about in this post. But give it another few years, and it will have pushed much more significantly beyond Emacs.

If you haven't tried it you don't know what you're missing. Either way, Emacs is going to need to adopt many TextMate features in the coming years, if it hopes to compete, a very difficult or even impossible feat, as it will mean breaking backwards-compatibility with existing Emacs infrastructure.

You mentioned having trouble setting up desktop.el. Here's my solution:

(require 'desktop)
(defun david-session-start ()
(interactive)
(david-kill-C-x-C-c)
(desktop-read)
(run-with-timer 0 120 'desktop-save "~/"))

(defun david-kill-C-x-C-c ()
(interactive)
(define-key global-map [(control x) (control c)] (lambda () (interactive) (message "No! Use the mouse!"))))

david-kill-C-x-C-c just saves me from fat-fingering C-x C-c and ending my emacs session.

I manually run M-x david-session-start on my One True Emacs (tm) session. That way when I run emacs on some random file from the terminal I don't have to worry about it loading my 8 million buffers from my .desktop file.

I completely agree with you about XEmacs. I used XEmacs for a long long time, but switched over to GNU Emacs at around pre-version 21 when it suddenly had almost everything that XEmacs had. My one missing feature is good tab completion in the M-x compile minibuffer. XEmacs has awesome tabbing which makes "make -C ~/src/some-random-directory" extremely pleasant. GNU Emacs still doesn't do this (by default).

There are some very interesting points in this post--interesting enough that you may have finally tipped me over the edge and convinced me to switch away from vim...to Yi.

Yi goes a step beyond Emacs or Firefox in that the entire editor is written in its "scripting language"--not a line of C. (Well, some non-core stuff does link to a few C libraries, such as gtk2, but those libraries could be replaced, though I wouldn't see much point in it.) It does this without performance penalty by not only embedding an interpreter in the editor, but a full compiler, and one that does pretty well in the performance shootouts.

I just need to find the time....

We work on XEmacs because it's more fun to work on this code than on other code. That's all there is to it.

Time to get back to havin' fun!

I am still a neophyte, but I love GNU-Emacs. If I could browse websites in it, I would (I know there are packages, but I am still too inexperienced to set them up). I love the buffers, dired, occur, and grep-find. I have written some simple ELISP stuff, and I like LISP, even tho it is difficult to approach sometimes.

I find that I spend most of my time in Emacs, and that time is more productive than time I spend in Visual Studio, or any other editor or IDE that I have tried, for that matter.

I think without ELISP, Emacs wouldn't be anywhere near as useful. I wish LISP was used more.

So what is the GNU Emacs equivalent of set-buffer-file-coding-system-for-read, anyway? That's the only thing I really miss.

Crashing? At least not on me but in some beta version and then also during some migration from Xfree to Xorgs X11 variant.

One of my xemacs is now alive for
> 180 days. It just sits around waiting for me to reconnect ;-)

The only thing that might get me away from XEmacs is another Emacs extensible with Common Lisp, Smalltalk, Ruby or Io.

I can not even stand the FSF tries any longer on selling varpoware for me. Since the 1999 or so the are talking about a Guile-Emacs, and where is it?

Another thing which might seduce me to try is a really good gnome or kde or the like integrated (X)Emacs

But this will not happen anytime soon and so I'm happy hacking away on my good 'ol XEmacs

Regards
Friedrich

Thanks Stevey, you have also inspired me to get cracking at improving rails-mode for emacs.

Great post. I wish this would happen. I am not a programmer but a long time emacs user. I do use a lot of the R and S programming language and a couple of years ago moved over to gnu emacs after over 6 years of xemacs. I am glad I did.
I just wish all the brains developing for emacs would get back under the same tent.

You mentioned self hosting development system and it reminded me of working in smalltalk/squeak. You write your code in the environment, compile and execute it as well as saving the state of all existing object instances.

Obviously everyone isn't beating down the door of the smalltalk/squeak crowd but Dan Ingalls created a similar environment in a browser using javascript and called it the Lively Kernel. I found this in the Weekly Squeak.

I would love to hear Steve comment on conkeror that was menitoned in a post above. I am curious about the prospects of that project.

Is there a Firefox extension that lets you add snippets of Javascript to your browser itself? Sort of like what greasemonkey does for individual webpages? That would be the main way to make Firefox more emacs-like....with Javascript. Extending it with C, the way most extensions do has one fatal flaw...a bad extension can crash the browser. You generally can't crash Emacs with bad Elisp.

I like Greasemonkey, but (last I tried it), it makes browsing slower...it seems to be linear with the number of GM scripts you bolt on.

For that matter, the original Netscape let you edit HTML pages...can't you do that in Mozilla as well? Making the browser into an HTML editor has been tried before.

For that matter, adding elisp scripting has been done several places...metacity (the GNOME window manager) comes to mind.

Lispworks still sells Liquid Common Lisp, née Lucid Common Lisp. The Lispworks Lisp product itself is not from Lucid.

Hey, Steve -

Great to hear your perspective. Much of this I share from a long history with Squeak.

We have just put together the makings of "another run at the fence" in JavaScript that might interest you. It's called The Lively Kernel, and it's a self-supporting system on a web page. We're just now getting many of the tools working (a new release due out next week), so you might be interested to take a peek.

Check it out (somewhat old version) at
http://research.sun.com/projects/lively/

I did some googling inspired by this post and found the most recent two posts at the following blog very interesting. Not only is the guy using emacs. firefox with the firemacs add on + conkeror but is also using a window manager configurable in common lisp which he does using slime from emacs

http://bc.tech.coop/blog/index.html

interesting setup.

"If Emacs can't find a way to evolve into (or merge with) Firefox, then Firefox or some other extensible browser is going to eclipse Emacs."

Dude, just because you can't figure out how to switch between windows on that weird Mac you picked up, don't make us smoosh all of our apps into one window^W app^W thingy^W whatever.

Please don't turn Emacs into a multi-language platform..

I agree with most of your other points, the world would be better of without XEmacs, and sharing backend with a popular browser (gecko, webkit...) would be an interesting path.

BTW: Gerd Moellmann is the main person responsible for dragging Emacs into the 90', he was maintainer and lead programmer for Emacs 21. He keeps a low profile on the net, so he doesn't get the recognition he deserves.

You make a good point about Emacs being self-hosting, but I don't think it's actually quite right. If Emacs were self-hosting, then the Lisp core would be written in Lisp. Last time I looked the core pieces of Emacs were still written in C.

Still, Emacs has some really attractive properties: it's extensible (of course) and it's reflective (or perhaps introspective) in that its extension language can inspect and modify itself. Or at least large pieces of itself. I think that's what makes Emacs really cool. Other systems, such as Smalltalk and the Lively Kernel, referenced above, share these properties.

By the way, I think it is possible for Java to be truly self-hosting. There is a SunLabs project Maxine to investigate exactly this. Unfortunately it doesn't look like they've published any reports yet, so I can't say much more. (I don't work on the project, but I have talked to the researchers.)

I hope they make elisp tail-recursive before emacs becomes some mutant cross-language platform thingy. I keep getting bit by the no tail-recursion thing , you'd think I would have learned by now.

Hi Steve,

I became aware of your blog maybe about a year ago, and have since read it periodically. More so regularly in recent months. Of all your writings, i almost agree whole heartedly, and the more i read, the more i see the years of experience and wisdom behind it.

However, there's something out of the ordinary today. You wrote:
«I'd also count among their legacies an absolutely outstanding collection of software essays called Patterns of Software, by Lucid's founder, Richard P. Gabriel. I go back and re-read them every year or so. They're that good.»

I'd consider that book one of the worst about computing i've ever read. More specifically, the first part of the book, i consider pure garbage, damaging to software community in a way similar to how religious people preachs about life-after-death or reincarnation or other troubling untruths. I wrote a review of this book in 1998: http://xahlee.org/PageTwo_dir/Personal_dir/bookReviewRichardGabriel.html

About Xemacs.... i heartily agree with you about its significance in the history of emacs. In my words, about maybe 4 years of commercial development (of xemacs), took the OpenSource/FreeSoftware camp 1.5 decade to catch up. For me, i started to use emacs in 1998, and from 1998-2005, i'm using xemacs 100% of the time purely for practical reasons. My experiences of using them is sum'd up here: http://xahlee.org/emacs/emacs_vs_xemacs.html

About your thesis that emacs is under threat of platforms such as modern browser... i think it is a very insightful point.

Making changes to emacs in some fundamental way wont be easy, not when RMS is ruling the party. Recently, his last email to me was “please don't be ridiculous”. All things considered, i think he's being somewhat rather rude. (For now, that was the last email between us. And, i think that effectly stopped me from wanting to contribute back to FSF on emacs, either of existing code or ideas)

It is relatively easy to contribute to FSF emacs by snippets of code, or a module. However, to change it in atypical way, such as modernize its documentation out of its 1980's context and writing in a systematic non-trivial update, or modernize its GUI out of the build up from 1980's ways, or, as you suggest, introducing or integrating or embracing some VM... am afraid will be nigh impossible, or take a long, long, long, time.

For me, i feel the most urgent change emacs needs is a few basic interface changes, detailed here: http://xahlee.org/emacs/modernization.html

The most effective method to make a stratedgic change, is, like Xemacs has done, or as Aquamacs has done, by simply having a capable person fork his own version with his vision, out in the public. If we assume that his idea is good, it will amass users quickly in a few short years, then, the GNU emacs will quickly adapt. This doesn't necessarily suggest dark politics or other conspiracy theories. It's just how the world works, by natural competition.

I didn't write this reply preplanned... but now at this point, i'm thinking, Steve: you are the one.

Xah
[email protected]
∑ http://xahlee.org/

☄

Re: aquamacs

I had it for less than 5 minutes on my computer. It had some weird behavior that I could live with. Something like, killing all the open buffers would close the application, instead of leaving a *scratch* buffer. Also, there may have been some bizzare frame (C-x 5 C-f) behavior. Anyways, it spooked me enough to make me realize that it's not *real* emacs.

Steve, we are Mozilla. You will be assimilated. Resistance is futile. Start coding that emacs extension for Firefox now, instead of losing time trying to resist. Conqueror is a good first step, but it needs improvements. You are the man for that.

;-)

You have to settle on some cross-platform-editor. Emacs is just an editor. But I can't stand RMS, that software socialist.

Therefore I use XEmacs, because it's much more liberal in my eyes. If XEmacs dissapears, I would stop using any kind of emacs.

>And you thought I'd given up on controversial blogs. Hah!

I for one hope you never do...
Great article!

The "ex" troglodyte speaks:

I don't want an editor to be configurable, extensible, self-hosting, or anything else. I want it to be the same everywhere, so that when I use it, I don't *use* it, I *am* it. When I type, it's not "I tell my fingers what to type", I just type. Likewise, when I edit with "ex" (after I had to give up Teco) I don't use "ex", I just edit. I no more notice what commands I am using, 99% of the time, than I notice what muscles I am using to type them.

The only problems I've had is that "vi" is usually Vim on modern systems, and all versions of Vim before 7.0 were abysmal implementations of "ex". To give you an idea, I had to make sure never to use the undo command, because it would undo everything you'd done since the last time you were in "vi" mode, which in my case was never! So everything got undone back to the point where I started to edit.

I now install the Heritage Toolkit "vi" everywhere I can't upgrade Vim or replace it with Nvi or the like. It will run nicely out of ~jcowan/bin, modulo the broken /etc/termcap, but terminal support doesn't add that much to "ex".

Don't think, though, that I don't appreciate working in a framework with near-infinite extensibility. I do; it's called the shell.

>> As soon as emacs gets proportional
>> anti-aliased fonts, I'll switch. I
>> can't go back to scratchy fixed width
>> fonts.

GNU Emacs 23 (wich is under active development and can be installed on debian/ubuntu systems as 'emacs-snapshot') does support proportional and anti-aliased vector fonts.

i almost forgot that....

http://fjl.wurzelzwergundmaerchenfee.de/img/euthanize.png

I tweeted a coupla wh00ts about UltraEdit and a buddy pointed me to this post.
Ok, so I don't use XEmacs.
But I gotta say: finding an article like this refreshes my belief that there's sentient life on Spaceship Earth.

wh00t wh000t!

What we need is hardware accelerated lisp machines, and a complete lisp os.

Dude, i've been using xemacs nearly every day since 1997 and i think it's crashed on me 1 time, maybe 2. My current session has been open since 5 or 6 days. Your article makes it sound as if xemacs is somehow unstable. Perhaps you were always working from the CVS Head?

Good job. I am eager for the unification.
As to the fundamental problems to solve, IMO, a better(maybe just little more modern and more fit in the whole desktop system each installation of Emacs inhabits) GUI is the most straightforward and effective way to gain more converts for Emacs, if it is not only geeks that we want to attract.

'm pulling a Yegge. Point is, use both. Use the dynamic side enough to find the static side ridiculous

I can't go back to scratchy fixed width fonts.

I have used XEmacs for 10 years now. I never recall it crashing, and I keep an XEmacs session open with half a hundred buffers for many, many months. I tried switching to Emacs lately (the new one with antialiased fonts), but it was so slow to refresh its windows when switching desktops that I gave up on it. The VNC session may be to blame, but XEmacs is less of a laggard in that respect, even with antialiased fonts.