Welcome readers! Herein lies a tale of woe. A tale of competing standards and compatibility. Yep, today we are talking about ANSI.
I’ve been working with the fantastic Enigma ½ BBS for a while now (and yes, I have to go grab that ½ unicode character every time I post something about it lol.) Enigma ½ is retro BBS software using a modern programming language (node.js) and technologies. It is almost more of a platform than a specific BBS software, incredibly flexible and extensible, with a wide range of supported protocols and functionality. As an example, one of the projects that is in progress currently is to enable ActivityPub integration so that it can talk to Mastodon and other fediverse systems. I’m not going to go into detail about that here, but you should check out more information about it on NuSkooler’s Blog - ActivityPub Beta. Lately, I’ve been working to reduce the backlog of issues on Enigma ½, particularly high priority ones. Specifically, right now I’m working on small incompatibilities between ANSI drawing programs, Enigma ½, and various terminals.
And oh-boy were there incompatibilities. I always knew that the standards were not always strict, but I didn’t realize how much the BBSes and terminals had to assume based on the loose standards, and that it made a real difference whether your art was drawn with TheDraw, ACiDDraw, or the newer PabloDraw, Moebius or upcoming IcyDraw, among many others. A partial list of ANSI editors can be found at: Ansi Editors. And then it makes a huge difference which client is used. Enigma ½ works with many different clients, from old terminals running on original hardware (Procomm / Procomm Plus for example, or even programs running on old 8-bit hardware,) all the way to brand new terminals like Alacritty, with many, many others in-between. We can broadly categorize them as:
- VT-100 terminals with UTF-8 character encoding
- VT-100 terminals with (some other) encoding
- ANSI terminals with CP-437 encoding
- ANSI terminals with (some other) encoding
- Something else completely
A word about Encoding⌗
What is this CP-437 and UTF-8? These are known as character encodings, and describe what byte or bytes make up a particular character. At the base is 7-bit ASCII, which is the base of most of the software we use. Here is a chart that shows the commonly used ASCII characters. There are variations on this, like ATASCII (Atari Ascii) and Petscii (Commodore Ascii) but for the most part this is the same on most of the systems we’ll be communicating with. Beyond 7-bit ASCII though there are differing ways of displaying extended characters, including the “semigraphics” characters used for drawing ANSI art. CP-437 is one such character set that became popular from early IBM PC’s and clones. A more modern encoding, UTF-8, is an encoding of the much larger and more complete Unicode standard, that sometimes takes two bytes to display a single character due to the number of available characters.
Enigma ½ internally uses UTF-8 for all of it’s processing, and converts into and out of CP-437, UTF-8 and ASCII (with more coming hopefully soon!) as needed. This is a great solution, as it means that we can support many terminals without code specific to each type. All we need to know is the mapping between the native encoding (such as CP-437) and UTF-8, and Enigma can then use either interchangeably.
The name “VT100” comes from a range of video terminals created by Digital Equipment Corporation. This line of terminals helped define the standards used up to today when sending control codes to/from a terminal client and server. These control codes could tell the terminal to clear the screen, move the cursor to a position, change text styles, and many more similar operations. Since these communications needed to be “in-band” - meaning that they share the same communications channel as the text communications, they use escape sequences to differentiate them normal content.
A related standard is (confusingly) known as ANSI. This terminology is confusing because while ANSI simply stands for the American National Standards Institute, what people are really talking about when they say ANSI in relation to BBS communications is the set of control codes that are mostly compatible with VT100 codes used in BBS software. A great resource for learning more about the ANSI BBS specification is at: (http://ansi-bbs.org). Information from the ANSI body itself can be found in their blog. The ANSI X3.64-1979 standard which it is based on was later withdrown in favor of an international standard, ISO/IEC 6429. So “ANSI terminal” is just a term now, better described in other documents - such as the ansi-bbs site.
ANSI and VT-100 are only mostly the same, and most of the differences can be glossed over. Enigma ½ takes the pattern of accepting input from either ANSI terminals or VT-100, and outputting only a subset that works on both known as bANSI (although, this also depends on what art files a sysop has created for their system.)
And what is this bANSI you ask? In 1995 (updated 1999) the author of a BBS client software named Bananacom, Paul Wheaton, authored an article called bansi.txt that spelled out how Banacom interpreted the various standards and common extensions to support a wide range of terminals. bANSI is what Enigma ½ and others use to help navigate the differing implementations of the existing standards and private extensions. Anyone looking to write a BBS terminal, software, or editor should definitely become familiar with this standard.
Back to the Project⌗
One of the first things that bothered me is that on some terminals there are only 23 rows available instead of 24 or 25. This happens, for example, because there is both a status line as well as a menu bar, each taking up a line. Even with the Enigma ½ screens set to 24 rows, with only 23 available the top row scrolls off the top. This shouldn’t be the biggest deal, because there is rarely anything really necessary on the top line, but we have a bigger problem. While Enigma ½ counts the rows for art files, it didn’t take into account that the art itself might scroll.
When the art scrolled, it would change the position of all of the text labels and other UI components, which would then display in the wrong positions. So the first thing I did was change the app to watch for instances where the row became greater than the terminal height, and perform a scroll function which moves the location of all of the UI elements appropriately. While I was at it, I added support for a few ANSI scrolling functions that were not previously handled.
The second issue came while I was testing for the first issue (as often happens.) When I was trying to save art files to display, I ran into a problem where all elements were appearing in the wrong places. After some debugging, I found that the culprit was an unhandled ANSI code that was being included by the art program Moebius that was not understood by Enigma ½. I took a spin back through the standard documents for ANSI, bANSI and v100, and added all of the unimpemented codes that I could find. This cleared up the positioning issues with Moebius, and probably a few others that we hadn’t run into yet. Hooray!
Next Steps and Conclusion⌗
While these changes have greatly improved the ANSI handling of Enigma ½, there are a number of areas that could use more work:
- The art system uses regular expressions to identify ANSI codes and other elements - this should probably be replaced with a state machine instead, which would be both more performant and standards compliant.
- Add support for other encodings. While we support CP437 and UTF-8, we should add support for plain-ASCII (there is some support in there already, but this should be expanded,) as well as other encodings such as PETSCII and ATASCII.
- Increase vt100 display support. There are additional DEC control codes that we could add support for, with functionality such as multiple pages, partial screen scrolling, and others.
- Add support for additional standards. I’d especially like to see RIP, Sixel, Videotex, among others.
As always, patches are appreciated, and please log any issues that you find, especially if you see anything that is still not working with the ANSI/vt100 support. I would love to see Enigma ½ have near-universal support for terminals. For more information about Enigma ½, see the website at Enigma ½ BBS or Github page