Cat's Eye Technologies: News Archive

A few more Befunge-93-related things that happened in 2018, the 25th anniversary year of Befunge-93. Since 2018 is rapidly coming to a close, this is our last opportunity to mention them in this context — and so here they are.

Silver Jubilee Edition of Befunge-93 released

Cat's Eye Technologies is pleased to announce the release of version 2.25 of the Befunge-93 distribution, christened the Silver Jubilee edition, commemorating as it does the 25th anniversary of the Befunge-93 programming language (1993-2018).

This release does not contain any changes to the code (those were all in 2.24, earlier this year,) but it does include significant changes to the distribution:

So how, you may well ask, do you obtain this humdinger of a release? Well,

ETERLAN SEPTEBMER and a minor revision of yoob

More on that Befunge celebration thing we're supposed to be doing this year!

For NaNoGenMo 2018 we constructed a novel generator called ETERLAN SEPTEBMER. Written in Befunge-93, it celebrates invention of Befunge in September 1993, a month called Eternal September in Usenet circles (for... other reasons.)

It generates a novel consisting of 50,000 really lovely words in short paragraphs. You can think of it as a kind of fireworks.

We have also released a minor revision of yoob, our framework for implementing esolang interpreters and visual debuggers. yoob is kind of considered deprecated anyway, since it was written as a Java applet, and support for Java applets is basically vanishing from web browsers.

But you can still use it, if you have Java installed on your computer. And you can still use it easily from a web page, if you have installed support for the Java Web Start (and I believe this comes with Java on many platforms.)

Version 0.3-2019.1129 of yoob makes this transition from Applet to Java Web Start official, and what's more, the Befunge-93 examples that ship with it have been updated — and they now include ETERLAN SEPTEBMER.

So check it out here!

Falderal 0.12, SixtyPical 0.17, and Samovar 0.2 released

A new version of Falderal, our literate-testing format intended for documenting programming languages, has been released. Version 0.12 is indebted to bug reports and patches kindly contributed by James Holderness. The reference implementation, py-falderal, also runs on Python 3 now.

A new version of SixtyPical, our heavily-statically-analyzed 6502-assembly-like programming language, has also been released. Now all symbol references can be forward references. The reference implmention also runs on Python 3 now, and has rudimentary support for producing programs that run on the Apple II.

Finally, a new version of Samovar, a DSL for describing simple simulation-worlds with propositions, has been released. We did this update as part of generating The League of Extraordinarily Dull Gentlemen for NaNoGenMo 2018. The reference implementation of Samovar also runs under Python 3 now.

Feedmark 0.6 and yastasoti 0.1 released

A new version of Feedmark has been released. There aren't many new features — in fact there are fewer features, as the entire archiving function has been offloaded to a new, separate tool called yastasoti — but there are new tests for existing features, and support for running the feedmark tool under Python 3.

Version 2.24 of Befunge-93 distribution released

As part of the 25th Anniversary of Befunge activity this year, the Befunge-93 distribution has now had its first significant release in 5 years: version 2.24.

This release includes many fixes and improvements, with contributions from Demur Rumed and James Holderness, to bef (the interpreter), bef2c (the compiler) and befprof (the profiler), and to the build system. The executables are now known to build on many platforms, including DICE C on AmigaDOS 1.3, Borland C++ 3.1 on FreeDOS, and MSVC++ on Windows.

This is the first of two releases that are planned for this year. The second release, fittingly numbered 2.25, will probably not contain any code changes, but will concentrate on documentation and example programs, and will include some interesting historical resources.

By the way, over at, Daniel Temkin has also written about Befunge on this its 25th year: Befunge (25 Years Of).

Funicular 0.5 and stringie 1.0 released

Funicular is our tool that's a bit like Vagrant, except instead of provisioning VMs, it provisions emulators.

Recently, we brought its Apple II support up to a working level, fixed its support for FreeDOS, and restored support for Amiga 500 which was removed entirely in a previous version because it was so lacking. We're pleased to report that it can now set up a development environment capable of compiling bef.c and several of our other projects under AmigaDOS 1.3.)

One such other project is our implementation of Underload in ANSI C, stringie. Vastly surprised last year that someone was actually using it, we decided to take it out of The Dipple and give it its own distribution, and put it on the path for a 1.0 release. It has reached that point.

Release of a new esolang: Equipage

We have designed and released a new esolang called Equipage. Equipage is a purely concatenative programming languages; it is what Carriage might've been in 2012, had I been less concerned about issues surrounding quoting at the time.

You can read more about it in the Equipage entry, or download the Equipage distribution for all the documentation and sources.

Befunge Turns 25 This Year!

Befunge turns 25 years old this year!

Cat's Eye Technologies has been planning to mark this occasion by doing some Befunge-related stuff in 2018.

We have a few things planned. I don't know how many we'll be able to get done, but they're planned.

Here are some we actually have done, so far:

Here are some we have planned:

ALPACA Version 1.1 Released

ALPACA version 1.1 has been released.

Version 1.1 clarifies a few edge cases in the spec, and introduces ALPACA Stylesheets 1.0, which can be used to specify the appearance of cellular automata.

Some of the internals of the reference implementation have been given an overhaul, and a few non-normative features have been added to it, such as being able to output SVGs (which can then be assembled into animated GIFs.)

More Pictures, Movies, and Gewgaws put Online

We've gone through some things we've done in the past few years and gotten them hosted online here at

We've put up a selection of Pictures we've made, mainly digital abominations using GIMP, but also a few pen-and-ink hand-drawn things too.

We've created an article curating the Movies we put together, mainly in 2017, mainly using kinoje and POV-Ray.

And finally, three Gewgaws that were long overdue, having been written in 2016, are now installed online:


2017 Year in Review

Here are some of our major accomplishments of 2017.

We messed around with POVRay again, after almost 10 years, and made some animated GIFs (which was new for me) with it, eventually putting together a script to orchestrate the process. Here are some of the generated movies:

We rebuilt the website; everything in the underlying database is written in structured Markdown now, making it much simpler to edit.

We started a place to write longer articles in, and wrote a couple that I'd been meaning to write for a long time:

We submitted 3 novel generators for NaNoGenMo 2017.

We made a few minor releases of SixtyPical:

The last of which makes it almost possible to write an actual game in it, and we're hopeful that we'll be able to release a 0.10 in early 2018 in which we bring the demo game up to parity with the demo game that was written for the first, inspired but unworthy version of SixtyPical, which got rebooted into the present project.

Finally, we should note that we've terminated our experiment with using Twitter as our social media presence. We have provisionally moved our announcements channel to on the open-source social network Mastodon.

Feedmark 0.3, Chrysoberyl, and

Version 0.3 of Feedmark has been released.

Several new features have been added, but the main progress that has been made is in its ability to rewrite the reference-style links in Markdown documents.

This allows us to build a "web" of Feedmark documents, where the entries are interlinked with Markdown link syntax, but the actual destination need not be given by the human writer, because it is determined by the system and automatically written in.

This, in turn, allowed us to rewrite all the data in Chrysoberyl as a set of Feedmark documents, which are converted to HTML5 very simply (Feedmark is a subset of Markdown, after all, and it's Markdown raison d'être to be transformed into HTML) to build the content of the website.

This, in turn, allowed us to jettison a bunch of old and grotty rendering code that was built on Jinja2 and YAML.

So, the practical result is that is largely a set of articles (such as Languages), and going forward, Chrysoberyl will be responsible for 2 orthogonal things: containing the underlying contents of these articles (e.g. and managing the distributions in which the implementations of the projects described in the articles can be found.

I hope to find this new format much easier to maintain and faster to update.

Release of version 0.4 of The Platform

On May 19, 2017, version 0.4 of The Cat's Eye Technologies Platform was released.

Since it is August already, this is an overdue announcement. Indeed, 0.3 was released on April 2, 2016, and that version was never announced at all — but we can gloss over that for the moment, as the improvements in that version were fairly modest.

But version 0.4, despite still being based on NetBSD 6.1.5, is quite a quantum leap, you see, quite a sea change. So it deserves an update, even if belated.

It goes like this.

After years of breaking it in, toolshelf was working well for me — I had grown quite used to eating my own dogfood — but it was failing in one spectacular way: it was making building The Platform kind of awkward. In order to run toolshelf, you need Python 2, which is not a part of NetBSD, so you need to build it first. This would not be a problem if toolshelf was already installed, you could just dock and build Python under toolshelf, I've done that before, but yes, you see the chicken-and-egg problem here, I don't have to spell it out for you.

So my solution was: rewrite toolshelf so that it can run immediately on a stock NetBSD install. That meant it would need to be written in something that comes with NetBSD. Which in turn meant either Lua or Bourne shell. After some debate, I picked Bourne shell. Developing software in Bourne shell is kind of brutal, compared to Lua, but on the other hand, it's quite well suited for dealing with filesystem and environment things, the kind of things that toolshelf does a lot of, and for which Lua, with its aim of portability and embeddability, is not very well suited.

The result is shelf. It is nowhere near as sophisticated as toolshelf was, but it does enough that I now use it regularly (and don't use toolshelf at all anymore) and enough for version 0.4 of The Platform to be based on it instead.

A similar thing happened with Funicular, the vaguely Vagrant-like tool that is used to build The Platform on top of a NetBSD system image in the first place. Funicular already was written in Lua, but, like toolshelf, Funicular does a lot of filesystem manipulation, and Lua is simply not as good at that as something like Bourne shell is, and what I had written I found to be poorly-structured and basically unattractive. So it, too, I re-wrote in Bourne shell, and cleaned up a bit in the process.

As for the actual contents of The Platform, the following new distributions were added to it since 0.2. These might have each their own belated announcement in the near future, but for now, here they are:

Many of these are also online, so you may find it worthwhile to follow some of those links.

The following distributions also suffered updates since 0.2, mostly very minor ones: ALPACA, BefOS, Braktif, Bubble Escape, Circute, Dungeons of Ekileugor, Etcha, Exanoke, Falderal, FBBI, Gemooy, Hev, Iphigeneia, Jaccia and Jacciata, Kosheri, noit o' mnain worb, Pail, Pixley, PL-{GOTO}.NET, Quylthulg, REDGREEN, Sixtypical, Super Wumpus Land, Tamsin, Thue (Cat's Eye Technologies' distribution), Whothm, Wierd (Cat's Eye Technologies' distribution), Wunnel, yoob.js and yoob and finally Zzrk.

You can download a pre-built image of The Platform from this torrent:

This feed has been rewritten in Feedmark

This feed has been rewritten in Feedmark. I'll get to what that means in a moment, but first I'll need to go through some history, for context.

Chrysoberyl was built to try to address two closely-related problems:

Before Chrysoberyl, the content of the website was split up into projects and "About Such-and-Such Topic" documents. I wanted something that put everything on the same level, and flexibly so.

So Chrysoberyl has "nodes" and each node can be anything and can have any properties (but some of them are statically checked.) And to take place of "About" documents, "commentary" can be attached to any node.

This worked well in many ways. But two things crept in over time:

That is, sometimes want to present information in a narrative context, but simply attaching commentary to nodes seems to do a poor job of that.

So I've begun presenting the information narratively, and extracting the curational information from that instead. This has been happening in The Dossier, which is a set of articles, written in Markdown, on various "curational" topics that didn't quite belong in Chrysoberyl.

But these writings are not entirely free-form prose. Embedded in the Markdown, there is metadata about the subjects being discussed. And since these subjects are specific, well-defined things, this metadata can be written in a machine-extractable form. And is. To this end, we created the Feedmark file format to define this embedded information.

And if we have a defined format, we can build tools to work with it.

And one of the things the reference implementation of Feedmark can do, so far, is generate an Atom XML feed from the metadata it has extracted from a Markdown file.

And in fact, that's what you're reading right now!

This wasn't entirely intentional, as Feedmark entries are not necessarily meant to be dated and ordered like a news feed is, but it makes sense when you think about it: Is a news feed not a sort of curational document?

And there are some nice benefits to it. As well as letting me remove stuff that doesn't quite belong in Chrysoberyl, it also makes it easier to update the newsfeed. Editing Markdown is easier than editing YAML, and various services (such as Github) can seamlessly format it into something passable (not spectacular, but certainly nicer than how they present YAML, and with virtually no effort.)

And since it is easier to edit, there is a greater chance of me actually updating it. Which I will try to do, again, soon.

Dropped support for Mercurial repos on Bitbucket

I like Mercurial. It was the first DVCS I learned to use, and I've always found it easier to use than git. I like Bitbucket too, largely because they'll give you private repositories for no charge, if you're a small team. I like that.

In fact, I like both of them enough to have gone to the trouble of keeping all of Cat's Eye Technologies' distributions mirrored in both git repositories on GitHub and Mercurial repositories on Bitbucket.

But alas, this has become far too much trouble in the upkeep, mainly because hg-git no longer works reliably for me. (You can see the fascinating bug report if you would like more details.) Even when it does work, it's yet another step I need to remember to take, and Mercurial doesn't work entirely reliably on my web server either now, etc., and basically I have better things to do with my time than hassle with this.

So, if you were using this Mercurial repositories on Bitbucket and are inconvenienced that you now must use Git repositories on GitHub instead, I apologize, but by all accounts you are in the distinct minority.

As for this news feed, I've been lax in updating it, and yes, I apologize for that too. I will try to post a catch-up update sometime in the near future.

RetroChallenge July 2015, and other updates

We've decided to enter RetroChallenge 2015/07. We're going to write some code... by hand! To this end, we've started a blog to track our progress, and have already made a few posts, even though it's not July yet.

Also, some new things done with the site since the last update:

Interview with Chris Pressey, by Daniel Temkin

A little while ago I was interviewed by Daniel Temkin on the subject of Esolang.

Since that's something I've been doing for ages, it goes into a lot of history, including pre-Internet stuff like BBS'es and the Amiga — so if you're interested in that, definitely read it!

And of course a bit of chat about Befunge, Gemooy, SMETANA, and such things, which may also be of interest.


In the meantime, I'll be over here, feeling old.

0x04 is the cruelest month: NaPoGenMo and altgamejam

April is National Poetry Writing Month! And as NaNoGenMo is to NaNoWriMo, so NaPoGenMo is to NaPoWriMo:

The Goal: Spend the month of April writing code that generates a poem. Repeat.

The Rules: The only rule is that you share at least one poem and also your source code at the end.

And who, I ask, can resist poetry written by computers?

For our part, we have written a few short poetry generators, mostly in the form of bash one-liners, whose output can be found in this Github issue.

More significantly, we've also written a Hello, world! program in the Beatnik programming language (and, for the record, determined that writing a quine in Beatnik is out of the question.)

Oh, but that's not all! Oh, but you'll wish it was all, you will!

For it is the case that the month of April is also when the ALTERNATIVE GAME JAM is taking place.

What exactly an "alt game" is we're still not sure, but that didn't stop us from making one. It's hosted on GameJolt as a condition of the jam, and there's this annoying ad that you have to wait through to see the actual game, but that shouldn't stop you from trying it. Here it is:

The New Gamerly Realism


Matchbox, a toy race-condition finder

We present Matchbox, a toy race-condition finder — it's an analyzer for a toy assembly-like language that runs in a web browser. It finds all interleavings between two programs, runs them all on shared memory, and declares there to be no race conditions if and only if every interleaving is either impossible, or produces the same result.

I wrote Matchbox to (cheesy as it sounds) raise awareness of race conditions. Also, Peterson's Algorithm is beautiful, and Matchbox graphically demonstrates that it does work. For a fuller account of why I wrote Matchbox, see the commentary on the Matchbox node, and the projects' README file.

Two more gewgaws have been installed online since the last announcement, too; Black Hole Poem and Noise to Signal No. 1. Enjoy.

One way to tile the plane with Wang tiles

tl;dr Backtracking Wang Tiler and it's pretty to watch go.

You're probably familiar with tesselations. Tilings, that is. A Cartesian grid tiles a plane with squares, a honeycomb tiles a plane with hexagons, and so forth. There are lots of examples. But the most interesting tilings, to me, are aperiodic ones — tilings that are not regular. Similar-looking sections of them may repeat, but the tiling as a whole never repeats exactly.

The most well-known example is probably Roger Penrose's kite-and-dart construction. But, as is often the case, the most famous example was not the first. The foundational work on aperiodic tilings was done years earlier, by Hao Wang, who came up with the concept, and posed the question (I'm paraphrasing):

Given a set of tiles, is there an algorithm that tells if they tile the plane aperiodically or not?

In other words, is the set of aperiodic tilings decidable? And Berger proved that it was not, by reducing it to the Halting Problem for Turing machines. Each aperiodic tiling can be put into one-to-one correspondence with a Turing machine that never halts.

That suggests that aperiodic tilings are tricky business, indeed.

But to be clear, this is a general question about arbitrary sets of tiles. Once you have picked a set of tiles, you may well prove it is aperiodic, and you may well tile the plane with them.

Of course, that business is still a little tricky. You can certainly work out an aperiodic tiling by hand, using human ingenuity and trial and error. Can you instruct a computer to do it? Yes, although unsurprisingly it leans a bit more on the trial and error side.

But wait — if a tiling corresponds to a Turing machine that never halts, how can you write a program for it? Ah, red herring — you just write a program that never quits. (Indeed, if you want to tile the "entire" infinite plane, what choice do you have?) But note that, technically, this might not be an algorithm, because algorithms are generally defined to terminate and produce a result. But we can put off that inevitable semantic debate by just avoiding that term.

At any rate, the process I've used is:

I'm not claiming this is efficient! It ain't, not by a long shot. But it's correct — it will tile the plane "eventually", as long as the set of tiles permit tiling (whether periodic or aperiodic) — and it can be pretty to watch go.

It's also apparent that "spiral" and "backtrack" are incidental; it's just nicely linear to arrange it this way, but you should be able to accrete random clumps too. Which might be even more inefficient, but even prettier to watch go. Dunno, would have to try it, and I haven't yet.

One slightly interesting thing about the implementation is that the backtracking is not implemented the obvious way, with recursion, because we want to display the state of the tiling as we go along, and Javascript in a web browser in particular won't allow that; it won't update the display until the current computation is finished and it's ready to handle a new event.

Therefore, backtracking is implemented with what are essentially continuations. All of the alternate, untried possibilities for each position are stored along with every tile, as it is placed. To be sure, this is not a general-purpose continuation, but it does encapsulate all the pertinent data for the tiling process — data that would otherwise be embedded in the function call stack, and not accessible to us.

The implementation is also in the public domain, so if you want to hack on it, feel free. It resides in the Backtracking Wang Tiler repository, in which you can also find further documentation on it. And maybe, in the future, other implementations and/or other tiling strategies.

Since this announcement has already gotten quite long, maybe I'll cap it off with an interesting question.

There are, as I understand it, Wang tilings which aren't even computable: they correspond to Turing machines for which there is no computable method to generate them. (They never halt, but we can't prove this.)

However, this automaton we got here operates randomly. The implementation is pseudo-random, to be true, but let's assume this is just an approximation of an idealized automaton with genuinely randomized tiling choices.

In a sense, it is always trying to find a non-terminating Turing machine. If the current tiling it's trying to lay down doesn't work (= represents a Turing machine which halts,) it backtracks and tries a different randomly-chosen tiling until it finds a a tiling that does work (= represents a Turing machine that doesn't halt.)

If left to run forever, might this automaton generate a non-computable never-halting Turing machine, in the guise of a Wang tiling?

Perhaps I misunderstand it, but I think it could. Just because you can't computationally generate such a Turing machine, doesn't mean that taking the limit of an infinite random sequence couldn't yield such a thing. Certainly, the closure of all such tilings would include these, and I can't see what's stopping this automaton from "accidentally" picking one.

Of course, taking the limit of an infinite random sequence could be a philosophically troubling concept — perhaps as philosophically troubling as generating an uncomputable object.

After all, if your tiling does tile the plane, you've proved the corresponding Turing machine does indeed never halt.

But it's hard to see how you'd know your tiling does tile the plane, without first knowing that the Turing machine does not halt.

Perhaps questions about infinity are best saved for people who have an awful lot of time on their hands.

Is Twitter your thing?

I ask because it is definitely not my thing.

But I recognize that it might be your thing, and that you might find it convenient to receive updates about Cat's Eye Technologies, via Twitter.

I mean, on a technical level, there's no reason it shouldn't serve as a good platform for announcements and updates. It's not my fault that that's not how most people use it.

And, hey, it's Friday the 13th today!. Spooky stuff happens on Friday the 13th, right? Spooky stuff like announcing that you are now on (ugh) Twitter.

Thus, I will grit my teeth and bear it.

Here, follow your little hearts out:

Content you can expect:

Frequency you can expect:

If Twitter is not your thing, fear not, as this self-same Article feed will continue, and will remain pretty much as it has been for the past, um, seven and a half years! (holy cow!)

Wierd (one of 'em) is now installed online

So back in the mists of time — by which I mean the late 90's — there was this three-way email conversation between me, Ben Olmstead, and John Colagioia, right?

And in it, we collectively sketched a fungeoid language which combined elements of Befunge-93 with elements of brainfuck and added an original element — it's the bends in the chain of symbols, rather than the symbols themselves, that determine which instructions are executed.

And now, a new page has been added to this language's long and fuzzy history: it, or rather, one of the extant dialects of it, has been implemented in Javascript and HTML5 with yoob.js and installed online here:

This may be the first implementation of (any) Wierd in a web browser. Or it may not; searching the web for "wierd javascript" doesn't return as many results as you might think (as, apparently, a lot of people think Javascript is pretty weird, and have a propensity to mis-spell the word "weird".)

I would like to implement the other extant dialect, too, some day. But not today.

Presenting Schrödinger's Game of Life

We present Schrödinger's Game of Life, which is exactly what it sounds like: Conway's Game of Life meets Schrödinger's Cat.

Each individual cell in the playfield may be in one of three states: Alive, Dead, or Possibly-Alive-Possibly-Dead (which we call Cat.)

Put another way, it's a cellular automaton that incorporates non-determinism.

To see it in action and play with it in your web browser, you can go over to its online installation.

For a full account of its development, see its README document.

Canvas Feedback: rated T for Trippy

Whether you get out the joss sticks and put Ravi Shankar on the stereo is up to you. But either way, you should check out Canvas Feedback.

It all started while discussing the 1970's-era Doctor Who title sequence and the video art of Nam June Paik. The digital era has made many of those analogue techniques less accessible — so, even if we can't bring them back, why not see what it's like if we try to adapt them for the modern world of web browsers and graphics cards?

This Maze is Never-Ending. Plus two gewgaws

A text adventure — yes, I could never quite get the hang of calling it "interactive fiction" — a text adventure game that I wrote in Inform about fifteen years ago, is now installed online as a Java applet thanks to Zplet.

If you like this, you might also like Cheshire Text.

If you hate this, you might instead like two new gewgaws that have also been put online: Eine kleine Glitchfraktal and Uncle Ankur. Neither one involves any text whatsoever (well, just a little on the control panels, maybe.)

Version 0.2 of The Platform - now with Torrent

Version 0.2 of The Cat's Eye Technologies Platform has been released. It's based on NetBSD 6.1.5 instead of 6.1.4, and has other minor version bumps like that in it. But the major thing now is that, if you don't want to go through the hassle of building it, you can download a pre-built image of it via this torrent right here:

For more information on The Platform, please refer to its repository on Github and/or the announcement of the release of version 0.1.

The Aftermath of NaNoGenMo 2014

So NaNoGenMo survived into a second year.

It probably would have made more sense to mention it at the beginning of November, rather than two-thirds through December, but — well — it's all very exciting when its time of year comes around you see, and I get distracted. Last year, I didn't bother to say anything about it even after it was over. So this is, at least, an improvement over that.

In 2014, Cat's Eye Technologies (or rather, Chris Pressey sponsored by the auspices of Cat's Eye Technologies — or something) staffed the NaNoGenLab, running one experiment in the generation, transformation, and general mutilation of text, per day, for the month of November. (So that was a goal of: 30 experiments. But we actually ended up with 32, somehow.)

Out of this chaos, we generated some novels too, which you can read in the Texts section.

And I regret with alarm to note that, right in the middle of November, there was an unfortunate... aesthetic incident... which resulted in... measures being taken, you see... resulting in the creation of this monstrosity. Listen at your own risk.

Oh! And, um, I dusted off some gewgaws that were just languishing in the lab, and put them on the site. They are, in no particular order, The Judgment of Paris, Woman on Film, Radialjective, and Circus Xamulus.

And also Cheshire Text, which is not a gewgaw but a text, but which was not done for NaNoGenMo, because it's not a novel. And a Text Uniquifier, which isn't a novel or a text but which was done for NaNoGenMo because it's a generator, or at least a filter, which I've classified, provisionally, as a gewgaw. Clear as mud, eh?

Introducing The Cat's Eye Technologies Platform

We are happy to announce the first release of The Cat's Eye Technologies Platform, version 0.1. It is a "distro" (if I may use that "word") of NetBSD 6.1.4, containing almost all of Cat's Eye Technologies' software distributions and the infrastructure needed to build and run them.

Each of the software distributions is tagged at a particular version (The Platform v0.1 can be thought of as a "versionset") at which they build and run on NetBSD, and if they have automated tests, the relevant ones pass.

There is no binary distribution yet, so if you want a copy of The Platform, you'll have to build it yourself (instructions for doing so are included.) However, we hope to distribute a binary image once we sort out how best to host large files.

It is, in a sense, an exercise in "software canning" to diminish bitrot. As long as you have an i386 emulator (such as QEMU) which can boot into NetBSD, you will be able to use these software distributions, without having to worry about conflicts due to upgrades and suchlike.

And as an exercise, it was somewhat illuminating; almost all of our projects, excluding what's hosted online, can be run on a rather modest environment of open-source infrastructure: NetBSD (which already contains Lua and gcc), GNU Make, cPython, Perl, Erlang/OTP (stripped down), Hugs98, Chicken Scheme, yasm, and Ophis.

BefOS version 0.10 released and installed online

Look! Up on this web page! Is it Befunge? Is it an operating system? No... no, it's just a disk sector editor with some unusual features, really.

Er... in case you were wondering... yes, this is the time of the year (September) when I usually do something to celebrate the anniversary of Befunge-93. It's 21 years old now, so if it were a person, it could drink legally in the United States, and, basically, it would be tried as an adult in pretty much any court in the world.

BefOS is a bit younger (but it would still be old enough to, what, drive?), but, in answer to your unspoken question, no, updating BefOS was not what I was planning to do for Befunge-93's birthday. It's only tangentially related.

In fact, I've been working on cleaning up the Befunge-93 (and Maentwrog and RUBE reference distributions — making their implementations compile cleanly on multiple platforms (Linux, NetBSD, FreeDOS, and AmigaDOS 1.3.)

But that's all not quite done yet; hopefully the Befunge-93 distribution will have a few more surprises in it before the end of the year.

But before you bemoan the barbarity of Befunge's birthday bash being a bit belated, ... well, nothing really, I just wanted to get some alliteration out of my system there. Have fun with BefOS.

Yolk, a Meta-circularly-defined Programming Language

And it came to pass that my mind turned once again to the question of whether there are any universal computational classes smaller than the class of recursively enumerable functions.

And lo, did I try to write a self-interpreter that was obviously not Turing-complete; and lo, did I fail. Pretty sure I failed, yeah.

But the result, behold! It was less than half the size of Pixley's meta-circular interpreter. So did I keep it, and did I christen it Yolk, and verily did I wax philosophical in its README about cons and stuff.

Release of toolshelf version 0.1-2014.0823

Oh man, so many things updated recently, but this news feed is so not one of them. OK, where to begin. OK...

Cat's Eye Technologies really isn't in the business of making tools. (Implementations of programming languages excepted — they're incidental, they're in service of the languages themselves.)

But we've implemented so many programming languages that, without some kind of tooling, the situation would become unwieldy. So, for example, we designed Falderal and build py-falderal so that all these programming languages could have their example programs presented nicely and tested in an automated way.

But there have been other, more fundamental problems to solve. How do you get all of these implementations onto a machine in the first place, for instance? If the distribution is self-contained, that's not a huge problem — download it and maybe build it and run it. But if it's not self-contained, if it has dependencies (and ultimately they all do, because each implementation is written in something,) then it becomes a package management problem.

And I detest package managers. I don't like their fragile package databases that get corrupted if you look at them wrong. I don't like their tendency to shotgun files all over my filesystem; nor do I like having faith that, if asked to remove the package, they will find them all and delete them. I don't like that they require root priviledges (root priviledges!) to install anything. And I don't like that they're almost never portable; there's one for every operating system and one for every programming language and never mind the fact that every modern operating system hosts more than one programming language.

So we wrote our own and called it toolshelf. toolshelf is a package installer which neither uses packages, nor installs anything. (Why yes, I am feeling a little farklempt, but that's neither here nor there.) It doesn't maintain a package database, it keeps all its files in one place (which you can just blow away if you don't want it anymore), it doesn't need root priviledges, and it is quite portable (in theory, anyway.)

It simply downloads (or clones) what you ask it to, makes its best guess at how to build it, and if that succeeds, makes its best guess at which files are executables and libraries that you care about, and puts those files on the appropriate search paths. Voilà! Sorted.

Ah, I know what you're thinking. You're thinking, that's insane. That's not a real package manager. That's going to fail in so many cases.

All true. And all irrelevant. In fact, it works in a surprising number of cases. It's actually really good for random little projects you find on Github and elsewhere. (It's not like they're ever going to show up in a "real" package manager. And when you get bored with them, you just blow away the clone directory.) And really, the other option is installing from source — and you can just think of toolshelf as an assistant that makes installing from source a lot less painful.

More to the point, it works really well with our projects. It is the recommended way to start using software from Cat's Eye Technologies. Just install toolshelf and then

toolshelf dock @@catseye

and watch it download, build, and install everything.

Even though it was originally only intended for random little projects, I now regularly dock things like QEMU, Chicken Scheme, and Django with it. I recently set up two Linux environments for my own use; on each of them, I installed only maybe a dozen packages from apt, and I haven't used pip at all. Most of what I've needed, I've brought in with toolshelf.

Of course, there are still some rough edges — and the way it works, there will always be rough edges — but there is a lot of potential too.

So, check it out!

Music, Pictures, Texts, and Wunnel

The Online Gallery has had an overhaul since the last time I announced anything regarding it. It's got tabs with proper anchors and stuff now.

In an attempt to prove that I am a polymath (which is a Greek word which means "bad at everything",) music I've written and pictures I've made and texts I've (meta-)written are available under the tabs there now.

But its main purpose is still for interactive HTML5-y stuff, especially esolangs, and lo, there is an HTML5 implementation of Wunnel there now too.

And with this implementation of Wunnel, yoob.js has reached another release point, version 0.6. But, of course, it is still basically in the work-in-progress stage, with things changing all over.

Github as a computer dating service (!)

As a little hack on the weekend, I started playing with the Github API, after using it for more serious purposes in toolshelf. The result was

Operation Match, Github edition

which turns Github into a computer dating service. Well, not really, and I really can't recommend using Github for romance...

...but the script does try to find Github users who have tastes that are similar to those to a given user, based on the starred repositories they have in common. And while it's very crude metric, I did end up finding a few users with it whose interests seem broadly similar to my own, who I have since started following.

(And I suppose that it's a coincidence that just as I was about to write this announcement, the BBC ran this article?)

De-pict-sly: Pixley and Etcha in HTML5

We present, just below, the Pixley reference interpreter (that is, the one that's written meta-circularly in Pixley itself) depicted as a set of nested, coloured rectangles. Looks a bit like an integrated circuit, perhaps, except without the wires. I'll leave it to you to decide if this is unsurprising or not. You can see this and more in Pixley.

Pixley Interpreter as nested rectangles

But that's not all! We also present what happens when you run that one Etcha example program over and over again. Looks a bit like a pattern produced by Langton's ant, and perhaps that is no surprise at all. You can try this, and any other Etcha program you care to concoct, yourself in Etcha.

An Etcha program state after many repetitions

Release of yoob.js version 0.5

yoob.js has been mentioned in several announcements, but this is the first announcement concerning it specifically. It started out as a project to replace yoob, running esolangs in HTML5 instead of a Java applet, but as of this release — version 0.5 — it has grown some generally-useful features for making animated and interactive things in HTML5, especially on the <canvas> element.

All that notwithstanding, this is a fairly minor announcement for a fairly minor release. The major version number is still at zero, and there will likely be a few backwards-incompatible changes in the next version. But 0.5 has enough reasonably useful junk in it that it's a comfortable snapshot of things at this stage.

Shelta v1.2 online in FreeDOS in v86

Version 1.2 of the Shelta distribution has been released. This is a very minor update, with the only change to the language being essentially inconsequential: Shelta source files now must end with everybody's favourite character, the vertical tab, instead of a null byte.

This was to work around what seems to be a subtle bug in FreeDOS. And that, in turn, was to get Shelta running under FreeDOS inside the HTML5-based IBM PC compatible emulator, v86. You can read all the gory details here.

Anyway, this all brings us to the good news, which is: you can now play with Shelta — including bootstrapping it — online, right here!

Presenting Amiga Gondola

This year — about two months ago, to be more exact — marked the 20th anniversary of Befunge-93. How to commemorate this historic occasion?

Well, one way would be to design its twenty-year successor, "Befunge-113". But that's just typical, y'know? Besides, I doubt there's much left to be done with Befunge besides suck all the fun out of it by defining it more formally. At any rate, I don't think any Befunge-113 I could come up with at the moment would make very many people happy, so: proposal declined.

A better idea: make it easier to re-create the development environment in which Befunge-93 was originally developed. Namely, AmigaDOS 1.3 on an Amiga 500.

So, that's what we did, and the result is Amiga Gondola, a set of shell scripts which, given an E-UAE emulator, images of the Workbench and Extras disks, and access to Aminet, put together an environment which, while not being a totally authentic copy of the environment I was using twenty years ago, is a perfectly cromulent development environment for AmigaDOS 1.3.

More information about Amiga Gondola and some nice screenshots can be found in its README.

Release of Falderal version 0.9 "Navy Pier"

Well, that was quick — well, we were on a roll (whoa, déjà vu.) Falderal 0.9 introduces one major new feature — the ability to supply input to a test. Typically, the test body would be considered the program (in some language) being tested, and the input would be sent to the standard input of the language interpreter (or whatever) to ensure that that program reacts the way you expect it to, given that input.

There are also a number of cleanups and minor improvements in py-falderal, including warning you if there are no tests to be run, or if it has run the same tests multiple times. It will also produce partial test results if you break it with ^C in the middle of a long test run.

Release of Falderal version 0.8 "Ukrainian Village"

A "new" version of Falderal, version 0.8 "Ukrainian Village", has been released.

I say "new" because, well, it's basically the same as it's been for the last year, when the Test.Falderal implementation of Falderal in Haskell was replaced by the py-falderal implementation in Python, which was christened the new reference implementation.

This switcheroo happened right after the last distribution of Falderal in which Test.Falderal was the reference implementation, version 0.7 "Odgen Avenue", was released. Unfortunately, I see by my records that I never announced that in this news feed, so, this may be news to you. Yes, it's written in Python now. And much like version 0.7 was a release of Falderal to get it out of the way so I could start working on a brand new code base (in Python), version 0.8 is a release to get this relatively-stable Python version out of the way so I can start making larger, long-overdue improvements to it.

The Falderal Literate Test Format has not changed much between 0.7 and 0.8. The main change is now that blocks must be indented by four spaces. It is generally assumed that Falderal test files will be embedded in Markdown and have file extensions like .markdown that let software (e.g. Github and Bitbucket) know how to pretty-format them.

In related news, recent improvements to toolshelf allow it to run any tests defined in a source tree, and to operate on all source trees owned by a given user on a given host. The end result being that we can now say toolshelf test and have it test all of Cat's Eye Technologies' active distributions. And the even better news is that, out of approximately 112 distributions, 22 have tests (mostly done with Falderal), and all 22 of those currently pass on both Ubuntu and Windows (under Cygwin).

New implementations of Xoomonk and Pixley

Xoomonk 1.0 was recently released. For this release, a Xoomonk interpreter was implemented in Python, using the Falderal tests written for the test-driven language design approach used in Xoomonk 0.1's documentation. I've used test-driven language design successfully in other projects (Castile, Robin) since Xoomonk was released, but it was the first project in which I proposed the concept, so it's nice to see it finally realized there.

The semantics of Xoomonk 1.0 are slightly simpler than those set out for Xoomonk 0.1 — specifically, a variable is considered unsaturated in a block if it is never assigned a value in that block (and not just if it is used in the block before it is assigned a value.) The syntax for built-ins also changed slightly. But these are not major changes, and the same original mouthfeel of malingering updateable stores is still there (and implemented!)

Pixley 2.0-2013.1024 has also been released. There are no changes to the language in this release, but two new implementations, mignon, in ANSI C, and haney, in Haskell, have been added to the reference distribution. They both implement Pixley well enough to serve as the bottom of a tower of Pixley interpreters. Many of the supporting materials (scripts, tests) have also been cleaned up.

I got two new gewgaws and an esolang... online

Two new HTML5 gewgaws in the Gallery: Chzrxl and Multicolouralism.

Also, the public-domain gewgaws are now kept in a repository on Github with a mirror on Bitbucket.

Also, the esolang ZOWIE is now runnable online. Unlike previous online esolang implementations, this one didn't involve re-implementing ZOWIE in Javascript. Instead, the ZOWIE reference interpreter, written in Python, is running under Skulpt on that web page. Only a few modifications were necessary, and it's only a little pokey.

Oh, also! The Programming Language index has been redone to be a bit more like the Project Matrix. It's a bit daunting, and awkward still, but probably better than how it was organized previously. hosted on new servers

"Here's a nickel, kid. Get a real web hosting service."

So I did.

You should see few if any changes. The ones you might see are:

Everything else should be pretty much as it was. If you see anything really broken, feel free to let me know.

Some HTML5 gewgaws for y'all

In a previous article, I mentioned that I consider, not just programming languages, but also HTML5, an artistic medium.

If short, fluffy, entertaining pieces of music are called bagatelles, then I propose that short fluffy works in the medium of HTML5 be called gewgaws. Or at least, that's what I call them.

And so, here are a few HTML5 gewgaws for your (possible) entertainment. Located in the Digital Art Exhibit of our Gallery. In fact there are six new ones in total. And when I say HTML5, I mainly mean the <canvas> stuff, of course.

They're not esolangs, but if you like esolangs, or if you liked Cyclobots, you might like some of the more mathematically-oriented ones.

I'd been meaning to re-implement this for ages

Turn back the clock. It's 1994. NAFTA. OJ Simpson. Rwanda. And SMETANA, implemented for the first time in Visual Basic.

And, unseen by all but maybe three or four people, Cyclobots.

Also implemented in Visual Basic. Described as an "interactive desktop toy". Fun to watch. Languished in obscurity because I had other interests at the time.

Then, the Visual Basic programs were forgotten, and eventually lost. When Java came on the scene, I had meant to re-implement it as an applet. But man, that must have been one low priority TODO item, because I never got around to it.

Until now.

'Course, it's not a Java applet now, it's Javascript on an HTML5 canvas, so make sure you're using one of those awesome new cutting-edge browsers, i.e., you've updated your Interwebs Explorer recently.

If you enjoyed watching what happens when you click "Revolution!" — and I mean really enjoyed it — feel free to show your appreciation by tossing a quarter into this here empty guitar case.

An Online Parade of Cellular Automata

It's about time! You can now play with Cat's Eye Technologies' cellular automata in your web browser on the following pages: REDGREEN, Circute, Braktif, Jaccia, and Jacciata.

I've wanted to be able to generate some kind of usable-in-a-web-browser code from ALPACA descriptions for a good long while. At one point I even considered hacking to generate yoob classes to do it. Well, the ALPACA language got a long-deserved update, and Perl gave way to Python for the new compiler; and yoob in Java has given way to yoob.js in Javascript.

(In fact, there's nothing much HTML5-specific about these Javascript implementations, except for a slider control for adjusting the animation speed, which is kind of crude and subject to change anyway.)

The few improvements to yoob.js that were spurred on by this effort to get these CA's online have also resulted in improvements to the online installations of Gemooy and noit o' mnain worb.

noit o' mnain worb implemented in HTML5

The implementation of various things in HTML5 continues. This time it's noit o' mnain worb, which you can play with here. With pretty graphics, for some value of pretty.

Seeing as worb has already been implemented in yoob, why do this? It's not like I can say it's "now playable online", because it already was. (Or if not "playable", then... "play withable"... or whatever.)

Well, the security track record for Java applets has been rather awful lately. Certainly, applets aren't going away, but they're quickly falling out of fashion. Updating your browser may, reasonably, disable Java. We may one day, not far off, see the JVM stop being shipped with browsers, and available only as an add-on.

And that all means I have to tell you to install and/or enable Java in your browser, which defeats the idea of reducing the install burden for esolangs, which was one of the original goals of yoob.

The newer yoob.js framework, which uses HTML5 instead of Java, is still pretty messy and ad-hoc, and lacks a few things that would be nice to have, like adjustable animation speed. But the more things, like this, that get implemented in it, the more it will mature.

Super Wumpus Land now playable online

Super Wumpus Land, our silly "extended dance mix" version of Hunt the Wumpus, has been converted to Javascript, and is now playable online here: Super Wumpus Land.

This new implementation uses modules from yoob.js to simulate a text-based terminal on an HTML5 <canvas> element.

Oops, I just turned Chrysoberyl into a blogging platform

I'm not much of a blogger. I've considered starting a blog a few times, but the thing is, most blogs are either about trends or opinions (or both, combined: reactions), and I don't follow trends, and when I have opinions, they're about specific things. (Which is why the old had an "About" section, and why Chrysoberyl lets commentary be attached to particular nodes.) And those opinions, like the things themselves, are relatively "evergreen" -- they're not news, there's no need to go announcing them. I also like to be able to edit/tweak/update an opinion incrementally after writing it, which also fits poorly with the blog model. [EDIT: see what I mean?]

At the same time, I do generate news, if not trends, and these announcements that I've implemented such-and-such or updated such-and-such are news articles, even if they're short ones.

And sometimes I have something to say that's not focussed on a single topic, and it fits best as a stand-alone article, such as my retrospective on esolang design.

Therefore, I merged the concepts of "News Item" and "Article" into just Article in Chrysoberyl, and all articles get included in the Atom feed, whether they're explicitly "news" or not. The result is that Chrysoberyl is, among all the other words that already don't describe it, now a blogging platform.

That's not to say you'll be seeing a lot of blog posts in the news feed. Even this article, which I consciously tried to write like a blog post, is still really mostly an announcement.

But you might see some. You totally might. Yeah, you're totally going to hear every little nitpick I have about why the latest version of Gadget X (resp. Language X) sucks and why GadgetCorp (resp. the Language X community) is going totally off-course as a company (resp. open-source project). Because that's so totally the kind of thing I give a crap about.

ALPACA turns 1.0

We have finally released version 1.0 of ALPACA, A Language for the Pithy Articulation of Cellular Automata.

The specification has been rewritten (although written would be more accurate) from scratch, and adds some new features to the language, such as being able to include an initial configuration in an ALPACA description, and handling neighbourhoods other than the Moore neighbourhood.

The reference implementation has also been rewritten from scratch, in Python this time. It can both interpret an ALPACA description, and compile it to Javascript. We plan to use the latter functionality to showcase some cellular automata on the website, which incidentally is looking a little less dire these days.

The version bump also cures two unpleasantnesses which are pet peeves of mine: ALPACA is no longer hanging out at version 0 (which sets it apart from, say, node.js), and the meaningless version numbers like 0.94 are now history. (It's not like it was the ninety-fourth iteration of anything.)

First language release of 2013: Exanoke

We have released Exanoke, a pure functional language which is syntactically restricted to expressing the primitive recursive functions.

If this doesn't sound like a stellar achievement, it's probably because it's not one. But, I was previously aware of only how an imperative language could be restricted to primitive recursion (i.e. have only for loops) and previously did some work on analyzing functions in a pure functional language to determine if they are primitive recursive or not; in a way, Exanoke combines both of those ideas.

Chrysoberyl is now in BETA

Happy 2013! Chrysoberyl is now in beta, whatever that means in this day and age where it's not considered absurd for web services to be in perpetual beta.

Well, concretely, what it means is we've replaced the older parts of []( entirely with Chrysoberyl entries, with a lot of web redirects so all those links out in the wild still work.

What it doesn't mean is that Chrysoberyl is finished in any sense. Its content will of course continue to evolve, and it will likely be a while before we are entirely happy with its structure and styling (so expect these to change in the near future.)

Language, Language, Implementation, Bug fix, Bug fix

So yeah, let's catch up with all the exciting, exciting non-Chrysoberyl-related developments we've seen here in the past few days!

First, please welcome our latest newcomer to the menagerie, Jolverine. I haven't been able to write a loop in it yet, but Ørjan is figuring that part out.

Second, I remembered why I designed Etcha — it was part of a larger, stranger, maze-based automaton called Zame. Which is now described on the esowiki.

Third, I finally got around to implementing Thue. In Ruby. I've been maintaining a distribution of it for so long without contributing to it, I felt it was overdue. And now there is a new release of the Thue distribution.

Fourth, Keymaker reported a bug with in — and while fixing it, I found another fixed that one too, and there has been another release of the SMITH distribution.

Lastly, while implementing a truth-machine in Ypsilax I found a bug in too (turns out, in Perl, the string '0' is false-y. It's things like that why I don't code in Perl much anymore.) So there's a new release of the Ypsilax distribution too.

Dungeons of Ekileugor: 3,221 bytes of roguelike

Have you ever wondered how much roguelike can be crammed into the 3583 bytes available to BASIC on an unexpanded Commodore VIC-20? Well wonder no more, for Dungeons of Ekileugor is such a cramming. Random dungeon level generation with fairly reasonable layout of rooms and tunnels, monsters, combat with hit points and experience points, queued messages, treasure, chests, traps, potions — it's quite a respectable set of "dungeon furniture" considering the platform, we think. Fire up your emulator (or, bless you, a real VIC-20) and try it out!

Chrysoberyl is now in ALPHA

Remember "The Great DVCS Exodus" (May 20, 2012)? (Well, if you don't, scroll down.) I am happy to say that it is, for all intents and purposes, complete: essentially all of our stuff is now in repositories on Bitbucket and GitHub.

And remember how I said that the role of might change in light of this, but that I wasn't sure how? Well, here's the deal. We have put together a database of sorts of all the things Cat's Eye Technologies has produced, and many of the things directly related to the things Cat's Eye Technologies has produced, and how they all relate, called Chrysoberyl.

I like to think of it as a sort of cross between a wiki and a semantic web and The Devil's Dictionary. It teases apart the relationships between programming languages (and other intangibles), implementations (things you can run), and distributions (things you can download), and casts it all in a bunch of YAML files with embedded Markdown. From these, a set of HTML5 pages can be generated, and lo and behold, these are now on the site.

They are, however, still in a very early stage of development. Expect the styling to change, the content to change, the structure to change, links to break, and the usual sort of thing.

This has a few implications for the rest of the site. The Gallery Space will stay more-or-less as it is for a while, but the Projects Space and the "About" documents will eventually go away, and will redirect into Chrysoberyl instead. In the meantime, they will likely not be updated to reflect the latest developments. Look at Chrysoberyl, or our repositories on Bitbucket and Github, instead, if you want to follow those.

And, more importantly for you (if you are reading this from our RSS feed) is that our RSS feed will no longer be updated. This will probably be the last item in it, except for an "Are you still reading this?" item when we've switched entirely over to Chrysoberyl. Instead, you will want to subscribe to our new Atom feed for news, generated from Chrysoberyl.

Cfluviurrh: an Esolang with Feeling (Literally)

We have released a new programming language. It's called Cfluviurrh. It was designed for writing programs that can feel. That is, defines a mechanism by which a program can be instructed to experience particular emotions. You might, thus, on first blush, consider Cfluviurrh to be unimplementable, as modern-day computers are not capable of experiencing emotions (you guess.) However, this is demonstrably untrue, and the Cfluviurrh reference interpreter demonstrates it.

The NEW Befunge-93 Reference Distribution

Since bef is the reference implementation of Befunge-93, or at least I've been calling it that, and since we've decided more or less that packaging all reference materials for a language into a "reference distribution" is the way to go, the bef and befunge93 distributions have been merged into a single, NEW distribution called "befunge-93". Also included are a couple of new example programs, a Makefile, and a small feature fix to the debugger. Enjoy!

One More Language: Hello, Velo

And when we do get rid of a language, we like to introduce another language to keep our numbers up. Enter Velo, a vaguely Ruby-inspired scripting language which unifies strings with code blocks, and scripts with object classes. It's by no means final or complete, but enough of it is there that it feels like it's been "invented", so here it is — expect it to change in minor (but backwards-incompatible) ways in the future, though.

One Less Language: Farewell 2iota

It's not often we get rid of a language here at Cat's Eye Technologies, but it happens occasionally. In fact, it happened today. I decided, some time back, that the languages beta-Juliet and 2iota were simply too similar to be distinct languages; and so, 2iota was absorbed into β-Juliet version 2.0 (which we can spell with the Greek letter now that everybody's got their Unicode boots on, right?) The semantics of the language have also been clarified somewhat (although the reference interpreter still needs to catch up to them in some small ways.)

yucca: Static Analysis for BASIC

We have officially released version 1.0 of yucca, a static analyzer for 8-bit BASIC programs. yucca can find those places in your program where you jump to a line number that doesn't exist, and a modest assortment of other things. It graciously ignores statements that it doesn't know anything about, allowing it to be dialect-agnostic. It has been used on our BASIC offerings (all two of them) and did, indeed, find a useless jump in Apple Befunge.

Bubble Escape now open-source

A new revision of the Bubble Escape distribution has been released. It contains all three versions of Bubble Escape: the 2009 Mini Game Competition winner (2K), an 8K cartridge image based on that source, and the original version, written in BASIC, from the 80's. What's more, the source code for all versions is now included, and placed under a BSD license.

The Great DVCS Exodus

Did you know that the website is statically generated with a set of XSLT templates? Well, it's true. This has certain advantages — like, when was the last time you saw broken XHTML on one of the project pages? But it has disadvantages, too. Namely, it's becoming a pain to update.

Meanwhile, we've been putting more and more of our projects under distributed version control. We tend to use Mercurial but hg-git provides a fairly seamless bridge to git repositories, too. And then there's these sites that host such repositories for you, Bitbucket and Github (among others).

So, we've been neglecting a bit lately, and have been putting our projects onto Bitbucket and GitHub, and making updates to them there. There are certain advantages to doing so. For example, if you find a bug in one of our implementations, you can fork that project, fix the bug, and even if Chris doesn't respond to your pull request for three or four years, at least your fork will be out there and available to anyone else who might be interested. On top of that, the source browsers on both of these sites are better than anything we have the time to invest in putting on this site. (Besides, if we go the way of _why, chances are some of our projects will end up there anyway, so why not give them a head start?)

But no, this does not mean the end of I mean, we have to keep it, or all our Java package hierarchies will be busted, right? And I think people are peeved enough over all the old links that are still around that lead to all the old domain names Cat's Eye Technologies has used in the past — no sense putting another one on that heap, if we can help it. But the role of might change somewhat. I don't know yet. We'll see.

Anyway, the point is: keep watching this space, but if you want breaking updates on what kind of crazy crap we're working on this minute, check out Cat's Eye Technologies on Bitbucket and/or Cat's Eye Technologies on Github.

Some Facts Regarding Apple Befunge v1.1

Pixley and Falderal. Falderal and Pixley. I know, I know, it's all you've been hearing about for the past three months, you poor thing. But now, O eso-pilgrim, well may you frolic and jubilate! I bring news of something closer to the weird. After more than a decade, some well-deserved updates have been made to Apple Befunge, that ol' "retrolanguage" for the Apple ][+, and it has been placed in the public domain.

I got Pixley power!

Pixley version 2.0 revision 2012.0219 has been released. There are no changes to the language, but there are a few significant improvements to the supporting tools and tests, and a few to the implementation, and a bevy of minor dialects of Pixley have been defined for purposes of Scheme-ly tar-pit research. Oh, and it's been officially placed under a BSD-style license now. Enjoy...

Release of Falderal version 0.6 "Streeterville"

The new year sees another release of Falderal, our framework for writing literate test suites for languages. This version supports variable expansion in functionality specifiers and the ability to add and remove functionality specifiers on the command line. It also fixes several bugs and shortcomings in the previous version. And, if you browse around the site a bit, you'll see we've started using it to format our projects' Falderal, Markdown, and Literate Haskell documents to XHTML.

Release of Falderal version 0.5 "The Loop"

Falderal, our format for unit-testing little languages, has seen release of version 0.5 "The Loop". As you can see, we have adopted a naming convention for release milestones — they're named after Chicagoland neigbourhoods, suburbs, landmarks, and institutions. This version was named after The Loop in recognition of its ability to shuttle test results between falderal and the various results generators implemented in different languages (Haskell, Bourne shell scripts) and to report on all the failures in a consistent way.

Release of Pixley version 2.0

Pixley version 2.0 has been released. It removes cadr and null? from the language. Partly because of this, the reference interpreter is now somewhat simpler: 124 lines of Pixley, with 407 instances of 53 unique symbols in 672 cons cells. The distribution now also includes driver shell scripts, Falderal tests, and a P-Normalizer, probably the first non-trivial Pixley program to be written outside the Pixley interpreter itself.

Madison, a Term-Rewriting Proof-Checker

One thing I've wanted to do for a long while is design a language in which one can state proofs of the properties of programs in that language. Not a full-blown theorem prover, just a proof checker, where you have to supply the proof, and the system tells you if it holds or doesn't hold. And not an immensely powerful proof-checker either, just powerful enough to state some simple proofs which hold over an infinite universe of values.

Well, after much thought and sketching, I have come up with a term-rewriting-based proof-checking language called Madison. It supports both direct proof and proof by structural induction, and I have used it to write a proof that the reflection of the reflection of any binary tree is the same as the original tree. It's not much, but I'm quite pleased with it.

Q: Are We Not Men? A: We Are Flobnar!

One day in September of 2011 — though I'm not sure precisely which one — marked Befunge-93's 18th birthday. That means that Befunge is now old enough to drink in its native land of Canada. To celebrate this, I thought I'd get Befunge-93 drunk to see what would happen.

What happened was Flobnar, an esolang which is in many respects a functional dual of Befunge-93; most of the symbols have analogous meanings, but execution proceeds in a much more dataflow-like fashion.

Release of Falderal version 0.4

Well, that was quick — well, we were on a roll. Falderal 0.4 introduces several desirable features. Test.Falderal is a Cabal package now, making installation easier, and this Cabal package installs a tool (unflinchingly called falderal) which drives the formatting and testing processes. Tests are now targetted at abstractions called functionalities, and functionalities are allowed to be implemented in different ways, including as shell commands, supporting the testing of multiple implementations of a function in multiple, essentially arbitrary implementation languages. This all brings Falderal closer to its goal of being a testing framework for programming languages.

Release of Falderal version 0.3

Remember that major change in how Falderal works that we talked about when Falderal 0.2 was released? Well, it's here, in Falderal 0.3. Basically, two things happened. One, Falderal is a file format now, and Test.Falderal is the reference implementation, in Haskell, for tools which claim to understand the format. Two, Test.Falderal is now able to format Falderal files into other formats. In fact, running tests is accomplished by formatting the Falderal file into Haskell code, and running that.


After years and years of apathy, we have fixed a number of bugs in the Flaming Bovine Befunge-98 Interpreter (FBBI), raising its quality level from "profound embarrassment" all the way to "marginal non-fail". Mainly we did this by randomly applying those patches that have been floating around, for which we owe much gratitude, but there was also a crippling memory-management bug that we found in the stack routines that highly deserved being killed, so, we sure did that thing too.

To celebrate — and to assuage our acquired annoyance at version numbers which look like decimal numbers that are supposed to indicate "close to release" — we have released version 1.0 of FBBI.

Note well that our general apathy towards Funge-98 has not abated in kind.

yoob 0.3 has been released, and it has moved out of Cat's Eye Technologies' lab and has become an official part of our website. It is now the main exhibit in the new Gallery of Interactive Esolangs. And, since we have three galleries now, we have developed this section of the website into something more real: our Gallery Space. This is where we will be exhibiting interactive works, and works not produced by Cat's Eye Technologies but which we find interesting nonetheless.

Have you seen LoUIE lately?

Have you seen LoUIE (our List of Unfinished Interesting Esolangs) lately? Because if you haven't, there's likely some designs there you haven't seen, because I just now added three (and in the past, haven't always announced it when I've added new entries.) There are twelve in total now, so if you're looking for an idea to build on, why not give it a look-see?

Researchers Discover the Civilization Advance "Language: Xoomonk"

Xoomonk is a programming language in which malingering updatable stores are first-class objects. Malingering updatable stores unify several language constructs, including procedure activations, named parameters, and object-like data structures. While the language is not yet implemented or even entirely finalized, it is pretty much complete, so is being released as version 0.1.

PL-{GOTO}.NET: Eat it Before it Eats You

You've always wanted an compiler for the example primitive recursive language PL-{GOTO} from Brainerd and Landweber's Theory of Computation, haven't you? And you have a need for it to generate MSIL which can be fed into ilasmto produce a .NET executable, don't you? And you wouldn't be satisfied unless it were implemented in Haskell, with a true-to-form grammar parsed with a Parsec combinator parser, would you?


Why aren't you answering me???

Bubble Escape 2K Now Playable Online

As subject. We've wanted to do this for a while, but last time an attempt was made, JaC64 was not quite up to the task. So we forked it on Bitbucket and for the past few weeks we've been busy fixing bugs in it. It's still not perfect, but it's playable on several platforms including Windows (Firefox and IE) and Ubuntu 11 (Firefox). So try it! Enjoy!

Release of Falderal version 0.2

Mainly to get a few minor niceties and bugfixes out of the way before we start on major changes to how Falderal works, Falderal 0.2 has been released. The planned major changes are probably more interesting than the niceties: I hope to make Falderal into something that can test more than just Haskell functions. Of course, it will still be written in Haskell.

First public release of yoob source code

yoob has pulled itself out of its "technology preview" stage, and is officially released as open-source software. More specifically, its source code is in the public domain and its development is hosted on a public repository on Bitbucket. Even though the source code is embarrasingly bad, I decided not to care. If you decide to care, you are quite free to hack it into better shape -- there are more than a dozen open issues in its issue tracker.

Corona: Realm of Magic: Dug out of the Attic

Something else I dug out of the attic. This was an elaborate roguelike I wrote (but never finished) in Perl, circa 2000. Still runs on a modern Perl (v5.10.1), but doesn't play too nicely with my modern terminal emulator. No further development is planned — it is retained here for historical interest only. It's actually been on the site for a while, but due to some technical problems, was not announced until now.

Apple Befunge: Dug out of the Attic

Hey, look what I dug out of the attic. I don't even remember what emulator the disk image was built for. Oh well.

Pail is an acceptable Bizaaro[sic]-Pixley

If you've been following our news, you've noticed that twenty-eleven's been kind of a light year for new languages here at Cat's Eye Technologies. It's been mostly updates, with the only original design being Wunnel. Well, that pattern's been broke! We have another new language. It started its life under the name Bizaaro[sic]-Pixley, but it's called Pail now (for PAIr Language).

Falderal: Literate Testing for Haskell Functions

We here at Cat's Eye Technologies have decided to stop building new ad-hoc test suite machinery for every new esolang we implement in Haskell. Instead, we have put together a package that can be used and reused for this purpose, and we have (for reasons obscure even to us) called it Falderal. It doesn't do a lot yet (it's only version 0.1...), but it has promise. We're already using it in Quylthulg. To encourage contributions, its development is hosted on a public repository on Bitbucket. Watch for its use in future projects!

Pluggin' some (specification) leaks in Eightebed

Eightebed version 1.1. Has a nice ring to it, doesn't it? Kind of mellow, with hints of strawberry and creosote on the back of the tongue. And that ring is even nicer when you realize that this update fixes the definition of the language to make it fulfill its original purpose. The concept of the safe area has been limited to statements in a block before the first free statement, thus preventing the creation of aliased dangling references. Thanks to Gregor Richards for pointing out this hole. Also included: a handful of bug and documentation of fixes that you don't care about.

Oozlybub and Murphy and Progress!

Version 1.1 of the Oozlybub and Murphy programming language has been released. This update tries to clarify some of the errors in the specification while also slathering some extra goo onto it like a wimpmode. Enjoy, then enjoy again, then PLEASE DO KEEP ENJOYING UNTIL YOUR EYES BURN WITH DEEP DEEP CORROSION.

More blowing off of dust: Maentwrog

Marinus has been so kind to write documentation and example programs for the ancient and venerable Maentwrog language that I decided I ought to update the distribution to include them — along with a small fix that allows the source to be build with modern C compilers like gcc and pcc.

yoob Grows Support for 3 More Esolangs

The yoob framework for esolang implementations has seen many minor improvements, and has had three more esolangs implemented for it. Two of these, Befunge-93 and brainfuck, are languages that no mega-eso-interpreter should be without; the third is Ypsilax, which was implemented in a relatively efficient manner (far more so than the existing implementation) to vet yoob's pattern-matching support.

None of these esolang implementations is particularly refined; each has some shortcomings w.r.t. the framework, and probably some bugs too. I expect they'll ripen with age. Enjoy...

Blowing the Dust off bef

The Befunge-93 reference interpreter, after a long period of inactivity, has been updated to version 2.22. This fixes three bugs, shortcomings, features, call them what you will: EOF characters are no longer written into the playfield when loading an otherwise blank source file (thanks to whoever sent me that patch six-odd years ago, and sorry that I have no recollection who you are); long source file lines are truncated instead of wrapping around to the next playfield line (this lets bef load mycology.b98 correctly); and the # instruction now behaves consistently when combined with wrapping (i.e. when it occurs at the very edge of the playfield.) All this and some minor aesthetic improvements to the interpreter, too!

Technology preview: yoob

Cat's Eye Technologies has embarked on an ambitious project: build a framework which both reduces the effort to implement esolangs, and makes those implementations easily accessible to, and usable by, anyone with a modern web browser — even non-programmers.

The project is called yoob. Even though it is still in an early stage of development, already more than ten esolangs have been implemented in it. And you can certainly expect more in the future. Enjoy...

We Three Things of Disorient Are

Alright, so it's a little late, I admit, but puns of that quality take time! We present three remotely fungeoid little esolangs that, in fact, were designed before Christmas: Gemooy, Nhohnhehr, and Kelxquoia. They're not in our project space yet, but they are on the esolang wiki, and one of them (Nhohnhehr) has an implementation already, thanks to Marinus.

So, take a gander at 'em, and you'll be saying, "Hey Chris, where'd you get your programming language designer's license? A cereal box?"

Actually, that'd be a pretty cool breakfast cereal, to have a prize like that...

Refurbishment of Thue Implementations

We've released a couple of bugfixes and some general modernization to the implementations of the Thue language included in our Thue distribution. Thanks to Nathan Thern for the bug report that spurred on this effort!

Fresh, Bold Stupid: Oozlybub and Murphy

Cat's Eye Technologies has developed a new programming language. The name of this language is Oozlybub and Murphy. Despite appearances, this name refers to a single language. The majority of the language is named Oozlybub. The fact that the language is not entirely named Oozlybub is named Murphy. Deal with it.

Pixley Version One Point One

An update to the Pixley programming language has been made, version 1.1, which fixes a problem where the reference implementation (and thus, for better or worse, the language definition) was not preserving the disjointedness of types. A lot of goodies have also been added to the distribution, including a more complete test suite, a REPL, a statistics generator, and an implementation in Mini-Scheme.

A Programming Language called Eightebed

I could have just explained to Gregor why he was mistaken, but noooo, I had to go and design an entire language to make my point. And I had to go and name it "Eightebed", too!

Whothm, a Language for Infinite Drawings

We present Whothm, a simple language for describing infinite two-colour bitmaps. You can try it out in your web browser with JWhothm, an implementation of Whothm in a Java applet.

Burro Climbs to 2.0

A major overhaul was done to the Burro language, resulting in version 2.0. Whatever Burro 1.0 might have achieved, it wasn't its primary goal. As Alex Smith was kind enough to point out, the set of Burro 1.0 programs don't actually form a group.

Burro 2.0 fixes the design of the language to avoid the problem. It is defined as an executable semantics written in Literate Haskell/Markdown which includes both a proof that Burro 2.0 programs form a group, and a demonstration of how one can map a Turing machine to a Burro program.

A video game: Bubble Escape 2K

Wow! A tiny C64 video game from the deep past! Actually, this was released almost a year ago, when it was submitted to the Mini Game Competition 2009, where it won first prize in its class. I decided to host it here given the undecided nature of the future of the site.

We humbly think you should get out your Commodore 64 emulator and play it!

At Cat's Eye Technologies, we have long beheld a few programming languages as exceptionally worth persual and appreciation, despite the fact that they have no implementations, are not under active development, and were not designed by Chris Pressey (or at least, you can't prove it.) We have, until recently, hosted these in our projects space. However, because of the criteria listed in the first sentence, we decided that these aren't really "projects" in any good sense, and we resolved to establish a distinguished display case for these beauties on this website.

And here it is – the Gallery of Esteemed Programming Languages.

The past few days have also seen a flurry of extremely minor updates to many of our projects, mostly to fix small conformancy issues, such as making the documentation validate as XHTML 1.0 Strict.

Another blast from the past: RUBE

In a move likely to set a dangerous precedent for retrotechnology maintenance, Cat's Eye Technologies today released a new version of the RUBE programming language, the first such update in over twelve years. "Actually 1.3 is exactly the same language as 1.02, but I couldn't stand that meaningless zero anymore," Chris Pressey, a spokesman for the company, stated at a fictional news conference. "At least I finally got around to getting the implementation to compile under something besides Borland C++."

None of Mr. Pressey's livestock were available for comment.

ZOWIE: Memory-Mapped Structured Flow Control

Cat's Eye Technologies' last language of the aughts, ZOWIE, goes to press. ZOWIE is a machine-like language, somewhat echoing SMITH in syntax, where flow control is both structured (as in structured programming) and memory-mapped (as in you write to memory to indicate the start, and the end, of each loop.)

Also, I can now say I've worked on a language project for every letter of the Roman alphabet. I'm so happy.

Release of Etcha, a Turtle-Based Language

We present the esolang Etcha, a four-instruction BitChanger descendant with a two-dimensional storage model based on turtle graphics. Unlike the turtle in LOGO however, the turtle in Etcha is an integral part of the computation, playing a role similar to the tape head of a Turing machine.

Dieter: Type Qualifiers meet Modules

After a long long time incubating, the Dieter programming language is released. Dieter (that's Dieter as in the German masculine given name Dieter, not dieter as in "one who diets") conflates type qualifiers with modules. The article describes how the interaction between these two features produces something that resembles object-oriented programming.

The Pixley Programming Language Arrives

We present Pixley, a strict subset of R5RS Scheme. Pixley supports four datatypes (boolean, cons cell, function, and symbol) and a dozen built-in symbols. The Pixley reference interpreter is highly meta-circular, being written in 140 lines of Pixley (or, if you prefer, 140 lines of Scheme.)

Pixley is also (depending on how you count them) my 50th programming language (that I'll admit to!) This puts me squarely in the ballpark of Wouter and Aaron, and suggests that I plan to be personally responsible for a significant fraction of the next 700 programming languages.

Scientific Proof that Cellular Automata are Intelligent!

Did you know that slime molds are intelligent because they can solve mazes? Well it's true, because a scientist said it! And now, since Cat's Eye Technologies has designed a pair of cellular automata (called Jaccia and Jacciata) that can solve mazes, we know that cellular automata are intelligent too! Three cheers for science!

The Unlikely Programming Language Unveiled

So we have our first new programming language of the year (or the overwhelming majority of it, anyway.) It's called Unlikely and it conflates objects with continuations, exposes its program structures as classes with commensurate inheritance relationships, and to top it all off, makes dependency injection mandatory. Overall a pretty painful experience, we think.

Shelta Revisited

Almost a decade after it was first published, the assembly-language version of the Shelta compiler has been translated to NASM. In the process it was improved so that it is both smaller than 512 bytes in size and able to participate in the bootstrap process. Check it out (if you like that sort of thing.)

A List of Unfinished Interesting Esolangs (LoUIE)

Wouldn't it be great if I had enough time to pursue every interesting idea for every yet-another-esolang I had? Well, that's simply not possible. They have leaked out into LoUIE, a List of Unfinished Interesting Esolangs, so that other esolang designers may possibly some day take up the torch. (And, considering their recidivist tendencies, probably commit arson with it. Just you watch.)

Nine Projects Moved to Archive

We've moved nine of our less exciting projects to an archive area of the website, where they can bit-rot in peace. Their distfiles are still available for download, but their project pages will not be maintained. A good number of these are forks and ports of open-source projects started by others (ErlGTK, ErlGuten, libvesa.)

Let's Have a Warm Hand for Quylthulg

Let's have a warm hand for Quylthulg, the latest atrocity to escape from Cat's Eye Technologies' labs. Quylthulg is a programming language with but a single control-flow construct: foreach. In fact, it does also have a goto, but that can only appear inside data structures.

Publishing of the "Kitsilano" Oscillator Circuit

After a summer hiatus, production resumes at Cat's Eye Technologies with the publishing of the schematic of and story behind 'Kitsilano', an electronic oscillator circuit based on a pair of NPN transistors and a single capacitor.

Release of the Context Rewriting Language Treacle

The Treacle programming language, successor to Arboretuum, has been released. It is based on context rewriting, which generalizes forest-rewriting; names and variables are separate, and patterns may contain holes inside which subpatterns may match at any depth.

Wee Present: The PETulant Cursor

Just in time for April Fools, Cat's Eye Technologies presents The PETulant Cursor, a tiny (just 44 bytes!) "display hack" for the Commodore 64. What's it do? Run it and see!

Arboretuum Forest-Rewriting Language Released

The Arboretuum programming language has been released. It is based on forest-rewriting, which, as the name suggests, is an extension of tree-rewriting in which multiple trees are rewritten simultaneously.

Website Updated to Use XHTML 1.0

We have updated our webpages to conform to the W3C Recommendation XHTML 1.0. (This does not, however, apply to HTML documentation in projects.) The CSS has also been cleaned up significantly, and the site generally looks better in Internet Explorer.

Release of the Larabee Programming Language

The Larabee programming language has been released. Larabee borrows the notion of branch prediction from computer architecture, and abuses it to create a state of total despair. Also great fun at parties.

Release of the Mascarpone Programming Language

The Mascarpone programming language has been released. Mascarpone is a rationalization and further exploration of some of the ideas behind Emmental. Mascarpone is a self-modifying language, defined by a meta-circular interpreter, in which interpreters are also first-class values.

Ypsilax Updated to Use Console::Virtual

Following the improvements made to the implementation of noit o' mnain worb, our Ypsilax implementation also uses Console::Virtual and optional sub-second delays to provide a nice screen-oriented animation of Ypsilax' two-dimensional, non-deterministic, reflective grid rewriting, taking the burden of visualization off the user.

RSS Feed Now Available for Cat's Eye Technologies

News on the latest developments at Cat's Eye Technologies is now available as an RSS feed. The news page is still available, but it is now automatically generated from the RSS feed by an XSLT stylesheet.

This may seem like a bit of a dodgy move, for a company with as staunch an attitude of post-modernist rectitude as Cat's Eye Technologies to go adopting technologies that are clearly of Pakled origin.

However, there are several good reasons. Firstly, XSLT, being a Turing-complete macro-expansion language with all the readability of Scheme, is practically an honourary esolang. Secondly, RSS is specified about as rigorously and consistently as most esolangs as well. And thirdly — although I dispute that this is a particularly important reason — somebody might actually find it useful.

Release of the Iphigeneia Programming Language

The Iphigeneia programming language is released. Iphigeneia is a toy programming language which contains features from both imperative programming and functional programming. It was originally intended as a testbed for algorithms that convert programs between the two forms, but it has strayed slightly from that goal.

Console::Virtual Revived, HUNTER and worb Benefit

Noticing that the Perl 5 implementation of HUNTER required a module that was never restored to the website after the last crash, I dug it out of cold storage and refurbished it a bit, resulting in Console::Virtual.

In the process I tidied up the HUNTER project quite a bit, including supporting a real delay, measured in milliseconds, between animation frames. (This requires the Time::HiRes module, but it still works without it; you just can't get sub-second resolution in that case.)

And, in the process of doing that, I noticed the implementation of noit o' mnain worb could use many of the same improvements. So now it, too, uses Console::Virtual instead of requiring an ANSI-compatible terminal, and supports an adjustable delay between frames.

Concurrent with this project interdependency, I've made a quick stab at listing the requirements for each project in the little "info box" on its project page. This is pretty crude right now, but it's hopefully a step in the right direction.

Release of the Didigm Reflective Cellular Automaton

Release of initial specification of the Didigm reflective ceullar automaton.

Also some random hacking on libvesa.

Release of the Emmental Programing Language

The Emmental programming language has been released. Emmental is a self-modifying programming language; it is defined in terms of a meta-circular interpreter, and this meta-circular interpreter provides operations that modify its behaviour. In fact, Emmental requires that this mechanism of meta-circular self-modification in order for it to achieve Turing-completeness.

Release of You are Reading the Name of this Esolang

The programming language You are Reading the Name of this Esolang was released. It's an exploration in the design space of programming languages with undecidable elements. Specifically, the problem of whether or not a given string of symbols is a well-formed You are Reading the Name of this Esolang program is undecidable.

Release of the Cabra Programming Language

The Cabra programming language, successor of sorts to Burro, has been released. Cabra programs form, not a group, but a dioid — an idempotent semiring — under the operations of sequential and parallel composition.

Release of the Burro Programming Language

The Burro programming language, after two years (on and off) of design work, has finally been released. Burro is a Brainfuck-like language whose programs form an algebraical group under the operation of concatenation (roughly speaking — see the docs for the complete picture.)

(Re-)Unearthing of the Maentwrog Programming Language

The Maentwrong language, predecessor of Befunge-93, and thought by me to be lost forever (again), was found (again) on a long-forgotten backup disc. It has been brought forth into the light of the projects directory (again) for whatever it's worth. No Longer in Service

The server is no longer in service This has two consequences: The old URL prefix for this site will no longer redirect here to, and the Subversion repositories served by will no longer be publicly available.

The redirect will continue to work, and the tarball releases of projects will still be available from

Whether the server will ever go up again or not depends on too many factors for me to be able to say at this time. I definately want to keep providing publicly available source code repositories of the projects, but due to circumstances it will have to be a low priority goal over the next few months.

Updates to SMITH and REDGREEN

The SMITH language has been updated in a tiny but significant way: overwriting instructions with other instructions is now defined. The reference implementation now implements this sanely as well. Thanks to Nathan Thern for pointing this out (and for submitting a SMITH version of "99 Bottles of Beer"!)

Some bugs in REDGREEN have been fixed as well: the documentation claims that Wires and Sparks behave per the WireWorld automaton, and that Zappy and BigZappy set things on fire. The ALPACA implementation of REDGREEN now properly implements these rules. Thanks to Stewart Gordon for pointing these bugs out.

Also, I dug up noise and put it in the projects. I swear there was a manual page for it too, but I can't find it.

Release of the Hev Programming Language

The Hev programming language has been released. Hev allows programming in infix notation, but at the same time, never needs parentheses and never forces you to memorize precedence tables! Truly, a major breakthrough.

Release of the Xigxag automaton

Xigxag, a simple automaton with exponential growth almost everywhere, has been released.

Zzrk Released, and More

Most significantly, a project has been added for Zzrk, a text adventure game written in a meta-language intended for building compilers (Zz).

I also brought GraNoLa/M and SP\ASM out of those dusty ol' boxes in the attic and added them to the line-up.

REDGREEN, Braktif, and Circute have also been split off from the main ALPACA distribution, and live in projects of their own.

I also polished the site design a wee bit.