TYPES / CLASSES / SOURCE FILES ------------------------------ cmd: Underlying implementation of command table management and command dispatching. Also handles the implementation of command aliases. commands: Dispatching for commands and handling of their arguments. (Commands are the interface exported to the user.) Many commands tend to be backed by functions. In general, a command takes a string as an argument and optionally returns a string. At the top of the file is a table mapping command names and help to functions implementing them. The standard entrypoint for executing a command is owl_function_command("foo"); Commands are only active within specific contexts, and attempts to call them from invalid contexts will fail. context: A context specifies the current state of owl, in terms of which modal window is active and which point in its life owl is in (eg, in startup, or running). This is implemented as a bitmask and there is some hierarchy. Commands may restrict themselves to only running in a limited number of contexts to prevent commands from being executed at points when they not make sense. Also, the data from the active context (eg, a pointer to an active window) may be passed to a command. dict: Simple dictionary abstraction mapping from strings to pointers. editwin: Text editing window (both multiline and single line). Sometimes also referred to as typewin. filter: Patterns which match messages. These may contain multiple filterelements which may be combined together (eg, by "and" and "or"). filterelement: An element of a filter which matches on some attribute of a message. fmtext: Formatted text routines (handles things like @i{foo}). These are particularly useful for building up text regions that are to be rendered on-screen, as they resize memory as needed, and they have routines for cropping as needed. functions: Where most features are implemented. Users should always interact with functions through commands. In general, functions are abstract entrypoints into the system and attempt to hide access to global state. global: Global state and variables and toplevel objects. owl.h defines "g" as a singleton instance of owl_global. Where possible/appropriate, most accesses to global data should be from a limited number of files (eg, from owl.c and functions.c). Consider whether you really need to before adding in uses of global. help: Help strings for commands and key bindings. Most of this is now in keys.c, commands.c, and variables.c, along with the definition of commands and keybindings. keys: Default key binding definitions for all keymaps. This also includes default actions for keymaps. keybinding: Binding between a sequence of keypresses and a command. When executed, this executes the commands. The sequence of keypresses is kept as a stack. Keybindings are a part of keymaps. keypress: Utility routines for translating between keypress values and and human-readable key names. keymap: Contains both keymap and keyhandler. A keymap is contains a list of keybindings, a sub-keymap, and optionally a default handler function. The sub-keymap is a more general keymap which is consulted if the specific keymap doesn't contain a match. (For example, the "global" keymap is the ancestor of all other keymaps.) The keyhandler is a collection of keymaps which handles checking for key matches within keymaps. It maintains a stack of keypresses and compares them against the bindings in keymaps. It also handles ESC as a prefix for Meta. At any one time, there is exactly one active keymap which determines where keybindings are looked for (along with its parents). list: Simple list abstraction. (Uses realloc to resize the list.) logging: Interface to incoming / outgoing zephyr logging. mainwin: Window that displays the list of messages. (Sometimes also referred to as recwin.) message: Abstraction to messages. Currently, messages are either of type zephyr or of type admin. messagelist: List of messages. owl.c: main() and signal handlers and other initial setup. Also contains the main loop, which is roughly: - handle scheduled resizes, and anything that might result - while zephyrs are pending, grab incoming zephyrs and handle them (which includes formatting them with either perl extension or default formatter as part of owl_message_create_from_zephyr). - updates mainwin display if there are new zephyrs - displays and updates popwins and the terminal as necessary - sends characters to the popwin, recwin/mainwin, or typewin/editwin owl.h: Prototypes for all types, as well as global constants. owl_prototypes.h: Autogenerated prototypes for all functions. Created by codelist.pl. popwin: Modal pop-up window container. Usually contains a viewwin for read-only scrolling text. readconfig: Perl extension interface. text: Text formatting utilities (ie, indenting, truncating, etc) util: Misc utility functions that don't fit anywhere yet: - sepbar rendering - tokenizing and parsing utilities - downstr - stristr - owl_malloc/free/realloc variable: Interface to setting and getting variables. Current variable types include bool, int, string, and other. There's also an enum type which is variant of int. Variables can be created and customized here as well. varstubs.c: Autogenerated headers for accessing global variables view: A collection of messages determined by a filter. Many operations may be performed on the members of a view, and a view can be narrowed-to for display. viewwin: Read-only scrolling text displayed in a modal popwin. This is also sometimes called "popless". zephyr: Routines for interfacing to zephyr. zwrite: Outgoing zephyrs. Sends pings on creation, handles command arguments, etc. =========================================================================== CURSES WINDOWS -------------- The four curses windows on the screen are recwin - receiving window sepwin - seperator window msgwin - message window typwin - typing window =========================================================================== Conventions and Design Criteria ------------------------------- Functions for creating and destroying objects should be named according to the following conventions: owl_*_init: Takes a pointer to a caller-allocated struct, and creates an object there. owl_*_cleanup: Takes a pointer, and destroys the object there, expecting the caller to free the struct. owl_*_new: Allocates a struct, creates an object there, and returns a pointer. owl_*_delete: Takes a pointer, destroys the object, and frees the struct. Functions should document if the caller needs to free something, and this should be the exception to the rule. Owl should be generally useful out-of-the-box without extensive configuration, for most people's needs. People shouldn't have to spend days tweaking with config files before being happy switching to it.