Next Previous Contents

7. Some Details on How Terminals Work

If you know almost nothing about terminals, it's suggested that you first read Introduction and also read Overview of How Terminals Work.

7.1 Terminal Memory Details

The terminal screen refreshes itself at perhaps 60 times per second from an image stored in the memory of the terminal. For a PC the monitor's image is stored on the video card inside the computer but for a terminal, the equivalent of the video card is inside the terminal. For a text terminal the storage of the image uses little memory. Instead of putting every dot (pixel) on the screen into memory and requiring the storage of about a quarter-million dots, a much more efficient method of storage is used.

A screen-full of text may be represented inside the terminal memory by ASCII bytes, one for each character on the screen. An entire screen only takes about 2K ASCII bytes. To display these characters, the terminal must also know the bit-map (the shape) of each of the almost 100 printable ASCII characters. With a bit-map of a character using say 15 bytes, only about 1.5K of memory is needed for the bit-maps of all the ASCII characters (the font). This ASCII text and font memory is scanned so that the resulting image is put on the screen about 60 times each second. This is a form of shared memory where a single bit-map of a letter such as the letter e, is shared by all of the many letter e's which appear on a screen-full of text. Low memory requirements meant low costs to produce monitors in the early 1980's when the cost of memory was several thousand times higher than it is today (costing then several dollars per kilobyte).

7.2 Early Terminals

The first terminals were something like remotely controlled typewriters which could only "display" (print on paper) the character stream sent to them from the computer. The earliest models were called Teletypes. The name "tty" is just an abbreviation for "Teletype". Early terminals could do a line feed and a carriage return just like a typewriter and ring a bell when a bell character was received. Due to the lack of significant capabilities this was the first type of terminal to be labeled "dumb". This type of terminal interface (using a terminal type called "dumb") is sometimes used today when the computer can't figure out what kind of a terminal it is communicating with.

7.3 Escape Sequences and Control Codes (intro)

Terminals have many capabilities some of which are always present and some of which require commands from the computer to change or activate. To exercise all these capabilities under the control of the computer requires that special codes be established so that the computer can tell the terminal what to do. There are two major type of such codes: escape sequences and control codes (control characters). There are many times more escape sequences than control codes.

Control codes

The control codes (or control characters) consist of the first 32 bytes of the ASCII alphabet. They include the following: carriage-return (cursor to far left), line-feed (cursor down one line), backspace, escape-character, tab, and bell. They do not normally show on the screen. There is usually a command which you may give to your terminal which will result in them being displayed when they are received by the terminal. It's called something like "Display Controls" or "Monitor". If you do this then the display may look a mess since escape sequences, which all start with the ESC (escape) control character, are no longer executed. Words which should appear at the top or bottom of the screen will show up in other locations. The escape sequences to reposition the cursor display on the screen but the cursor doesn't move to where the escape sequence says.

Escape sequences

Since there are not nearly enough control codes to do everything (and for some reason, not all of them are utilized) many escape sequences are used. They consist of the "escape" (ESC) control character followed by a sequence of ordinary characters. Upon receiving an escape character, the terminal examines the characters following it so that it may interpret the sequence and carry out the intended command from the computer. Once it recognizes the end of a valid sequence, further characters received just display on the screen (unless they are control codes or more escape sequences). Some escape sequences may take parameters (or arguments) such as the coordinates on the screen to move the cursor to. The parameters become a part of the escape sequence. An Esc Sequence List is on the web for some terminals, but it's terse.

A list of the escape sequences for your terminal should be in the "programmers manual" for the terminal. Except for very old terminals, there may be two or three hundred such sequences. If you don't have a such manual it's not easy to find them. Some of the sequences are available on the Internet. One link is Esc Sequence List. By searching the Internet for one sequence (such as ESC[5m) you may come across a long list of them.

Another way to determine some of them is to find the terminfo entry (termcap) for the terminal and mentally decode it. See Terminfo and Termcap (detailed) in this document and/or the Termcap Manual on the Internet. Unfortunately, the terminfo (termcap) for a terminal often does not list all of the escape sequences which the terminal has available for use, but fortunately, the most important ones are usually there.

7.4 Display Attributes & Magic Cookies

Terminals have various methods of generating character attributes such as bold, reverse-video, underlining, etc. There should be no need for the user to worry about how how this is done, except that it creates problems for some old terminals and there is sometimes an option for this in the set-up menu of newer terminals.

The magic cookie method is obsolete. It's the simplest (and worst) method of defining attributes: Use a certain byte for the start of an attribute and another to end that attribute. For example, a "start underlining" magic cookie byte is placed just before the first word to be underlined. These extra bytes are put into the memory of the screen page, just like character bytes that display as characters. But this might foul up the count of the number of characters per line since non-printable magic cookie characters are intermingled with other printable characters. This sometimes causes problems.

A better method which uses more memory is to assign an attribute byte (or half=byte, etc.) to each displayed character. This method is used by PC video cards (for text) for the common PC monitor.


Next Previous Contents