ComplexFunctions

Introduction

A Complex Function in FVWM is nothing more than a series of commands grouped together to perform a given task. How that function operates depends entirely on how it was written and what [[FunctionContext]] it runs in. A function can have any number of commands, including builtin FVWM commands, external commands, etc.

Declarations of Functions

The formal declaration of a function is started with the AddToFunc keyword, its syntax being:

AddToFunc [name [I | M | C | H | D action]]

To take an example, suppose we wanted to write a function (a useless one in this context, but serves for demonstrative purposes) called !RaiseMe which raises a window as soon as a function is called. We would write:

AddToFunc RaiseMe I Raise

We might decide we want to bind that function to the titlebar of a window:

Mouse 1 T A RaiseMe

Such that when we then clicked on the titlebar with mouse button 1, the window is raised. Now let’s assume in addition to that we wanted to (at the same time), move the window to the previous page as well.

Since we have already declared the function !RaiseMe earlier on, we could continue to add to its definition as in:

AddToFunc RaiseMe I MoveToPage prev

The point being here is that each time a new definition is added to the function, the AddToFunc keyword does that, although it can be an issue at times since it’s somewhat harder to follow. There is another way of declaring functions, and that is to list each operation of it below a single AddToFunc definition using the group operator +.

AddToFunc RaiseMe
+ I Raise
+ I MoveToPage prev

The plus operator just extends the definition of AddToFunc (but works in the same way with AddToDecor and AddToMenu), effectively grouping the elements after it together.

Removal of Functions

It would be quite annoying if we could only ever continue adding to a function without somehow being able to remove its definition. Luckily though, FVWM has the DestroyFunc command to do this:

DestroyFunc [name]

Hence:

DestroyFunc RaiseMe

…would make FVWM forget the definition of that function.

Triggering actions from Functions

The astute will have noticed that the first parameter to a function is a single letter. Those letters define what part of the function runs, depending on certain conditions, hence:

Operator Context Meaning
I Immediate Called when the function is invoked.
M Motion Called when an item is moved.
C Click Called when an item is clicked.
D Double-click Called when an item is double-clicked.
H Hold Called when an the pointer is held on an item.

Usually (although not always) “I” is used the most often, as it denotes something to be done regardless of whether one is clicking on something, or wanting to move an item, for instance.

“M” is useful if you want to have something happen when a window is moved. This action is triggered as soon as the window starts to move, and can sometimes conflict with “H” if it is used within the same function. For example, “M” is useful to, well, move a window, since it might not be an implicit operation. “H” tends to be designated the same value as “M” (and vice-versa) when they’re used within the same function as it often then means there’s no ambiguity in the delay between the window moving, and the button being held within the window.

“C” and “D” are useful to define actions for clicking on a window, or parts of it.

Here’s an example: FuncMoveOrRaiseLower

DestroyFunc FuncMoveOrRaiseLower
AddToFunc   FuncMoveOrRaiseLower
+ H Move
+ M Move
+ C RaiseLower

… which says that when the function is called, if the mouse button is either held or the window is invoked to be moved by the mouse button, then the window moves, otherwise if there’s a single-click instead, the then the item is either Raised or Lowered, depending on what state it is in before the click event occurs.

Piecing the parts together

When writing functions it should perhaps be obvious then that removing a function’s definition before you define it (irrespective of whether it is the fist time the function has been defined) is probably a good idea, hence:

DestroyFunc myFunction
AddToFunc   myFunction
+ I ....
+ M ....

There is no agreed definition as to function naming conventions, other than to choose something and try and be consistent in its use. I (ThomasAdam) like to use the prefix ‘‘Func’’ and then a descriptive name about the function. Like most things in FVWM, the actual name of the function is read case-insensitive to FVWM although the use of !CamelCase as a convention is quite popular.

Function Inputs

When calling complex functions you can provide the function inputs as a space separated list. Each input is stored as a variable $0, $1, $2, …, $9. Use $[n] to access a parameter bigger than 9. Additionally you can use $[n-m], $[n-], $* or $[*] for different collections of the input(s). For example

# Function: MoveClickX $0 $1 $2
#   $0 - Action on Mouse Hold + Move
#   $1 - Action on Mouse Click
#   $2 - Action on Mouse DoubleClick
#  Example: Mouse 1 T A MoveClickX Move Raise Maximize
DestroyFunc MoveClickX
AddToFunc MoveClickX
+ M $0
+ C $1
+ D $2

This function can then be bound to window buttons, title bars, etc like

# Titlebar: Move or Raise on single click or Maximze on double
Mouse 1 T A MoveClickX Move Raise Maximize

Function Caveats

I’m sure by now functions sound great – and they are! But like everything, there’s a few caveats to bear in mind with them. During the lifetime of a function (that is from the time it was invoked to the time it ends), the pointer is grabbed. What that means is that the mouse pointer is under control of the XServer until it is released. For most things this won’t matter much, and the user (that’s you) won’t be aware of it. However there is some programs (such as ‘‘import’’) which need the pointer to actively workout what screenshot it is going to take. This won’t therefore work the way it’s intended from within a function such as:

+ I Exec exec import -window root ...

To alleviate such problems, one would have to use PipeRead to fork off a new process (away from the function).

Synchronisation is another important issue within function execution. To take an example, suppose you have the following function that performs a series of operations:

DestroyFunc FuncSomeFunction
AddToFunc   FuncSomeFunction
+ I Raise
+ I AnimatedMove 500 300
+ I Lower

You might expect these commands to run one after the other. They do, although there are times when that might not always be the case. FVWM performs no FunctionSynchronisation as to the ordering of the commands – or rather it makes no guarantees. If a strict set of synchronised commands is needed, it is recommended to use PipeRead – since FVWM will wait for the PipeRead to return before carrying on with the rest of its input (such as the other commands in the function).