If MicroEMACS has been built with Ruby support, it will also support the notion of modes, which are similar to major modes in Emacs. A mode consists of a name (which is arbitrary) and a set of key bindings local to that mode, and is attached to a specific buffer. By default, buffers in MicroEMACS do not have modes, but you can provide support for modes by loading Ruby support for them. Modes can be used to implement features for particular types of code, or to provide special types of buffers that are not treated solely as plain text.

Mode initialization

When MicroEMACS reads a file into a buffer, or visits a file (which may or may not exist yet), it calls the Ruby function initmode, which lives in the file pe.rb, and which gets loaded whenever MicroEMACS starts. This function attempts to determine the name of the mode associated with the file being edited. First, it examines the first few lines in the file itself for a line containing a string in the following format:

-*-MODE-*-

where MODE is the name of the mode associated with the file.

If such a string cannot be found, initmode then examines the name of the file itself, and attempts to find a match in the $modetable array in pe.rb. This table associates filename patterns with mode names. The table has very few entries by default, but you can expand it as necessary by editing pe.rb directly, or by overwriting it or adding to it in your own Ruby extension.

If initmode can determine the mode name, it calls the function MODE_mode, where MODE is the name of the mode. For example, if initmode determines that the mode name is c, it calls the function c_mode, if it exists. This function is called the mode hook, and it is responsible for performing all necessary initialization for the mode. If the mode hook does not exist, MicroEMACS does not report an error.

Simple mode example

To see how a rudimentary mode is written, look at the file mode.rb in the MicroEMACS source code. (This file should also have been installed in /usr/local/share/pe, if you built MicroEMACS and ran make install.)

Here is the code for c_mode, the hook for C files, mentioned above:

def c_mode
  setmode "C"
  bind "gnu_indent", ctrl('j'), true
  echo "This is C mode"
end

The first line in this function creates a new mode for the current buffer, and assigns it the name "C". The name is arbitrary, and is used only for displaying in the mode line for the buffer. But setmode must be called; otherwise a mode will not exist for the current buffer, and attempts to mode-local key bindings will not succeed.

The second line in this function binds the MicroEMACS function gnu-indent to the C-J key. This binding is local to this mode only. If you switch to a buffer that has a different mode, or no mode, the binding for C-J may well be different.

The third line in this function is present only for debugging purposes. It can be safely deleted.

You can load this rudimentary mode support automatically by adding the following line to ~/pe.rb or ./pe.rb:

load 'mode.rb'

A more complicated mode example

The file dired.rb, also provided with MicroEMACS, is a more elaborate example of a mode. It implements a directory browser similar to the dired mode in Emacs. Unlike the simple example shown above, it provides both global and mode-specific key bindings, and it does not provide a mode hook function. Instead, it provides a new command, dired, that is globally bound to the keystroke C-X D:

ruby_command "dired"
bind "dired", ctlx('d')

The file dired.rb also creates three new commands but does not bind them to keystrokes immediately:

ruby_command "visitfile"
ruby_command "openfile"
ruby_command "displayfile"

The initialization of the mode happens in the dired function when you enter the keystroke C-X D. This function prompts you for a directory name, then calls showdir to open a view on the directory.

The showdir function opens a new window with a buffer called *dired*, to avoid conflict with existing buffers. It clears any existing contents of the buffer. It then runs /bin/ls -laF to load the buffer with a directory listing. The first line in the buffer contains the directory name, and each subsequent line contains information about a file in that directory, as provided by ls. If the name of a file ends in a '/' character, that file is actually a subdirectory. Finally, showdir marks the buffer as read-only.

The showdir function then performs some string matching to determine that starting column for filenames in the directory listings. Finally, it creates a mode for the directory listing and attaches three key bindings to it:

setmode "dired"
bind "visitfile", ctrl('m'), true
bind "openfile", key('o'), true
bind "displayfile", ctrl('o'), true

The bind calls create key bindings for the three new commands that were defined earlier. These bindings perform three distinct actions on the file under the cursor (called the "selected" file):

• The Enter key opens the selected file in a new window, replacing the current dired window (which still exists).

• The o key splits the screen into two windows, one containing the dired buffer, and the other containing the selected file. It then moves the cursor to the selected file.

• The C-O key is similar to the o key, except that it does not move the cursor to the selected file.

You can load the dired mode support automatically by adding the following line to ~/pe.rb or ./pe.rb:

load 'dired.rb'