4. Reviewing the basics

G was a gamester, who had but ill-luck;
H was a hunter, and hunted a buck.
../_images/picG.png

oing through the design of our first game in the previous chapter has introduced all sorts of Inform concepts, often without giving you much detail about what’s been happening. So let’s review some of what we’ve learnt so far, in a slightly more organised fashion. We’ll talk about Constants and variables, Object definitions, Object relationships – the object tree, Things in quotes and Routines and statements.

Constants and variables

Superficially similar, constants and variables are actually very different beasts.

Constants

A constant is a name to which a value is given once and once only; you can’t later use that name to stand for a different value. Think of it as a stone tablet on which you carve a number: a carving can’t be undone, so that you see the same number every time you look at the stone.

So far, we’ve seen a Constant being set up with its value as a string of characters:

Constant Story "Heidi";

and as a number:

Constant MAX_CARRIED 1;

Those two examples represent the most common ways in which constants are used in Inform.

Variables

A variable is a name to which a value is given, but that value can be changed to a different one at any time. Think of it as a blackboard on which you mark a number in chalk: whenever you need to, just wipe the board and write up a new number.

We haven’t set up any variables of our own yet, though we’ve used a couple which the library created like this:

Global location;
Global deadflag;

The value of a global variable created in this way is initially 0, but you can change it at any time. For example, we used the statement:

location = before_cottage;

to reset the value of the location variable to the before_cottage object, and we wrote:

if (nest in branch) deadflag = 2;

to reset the value of the deadflag variable to 2.

Later, we’ll talk about the local variable (see Routines) and about using object properties as variables (see Objects).

Object definitions

The most important information you should have gleaned from the previous chapter is that your entire game is defined as a series of objects. Each room is an object, each item that the player sees and touches is an object; indeed the player herself is also an object (one that’s automatically defined by the library).

The general model of an object definition looks like this:

Object obj_id "external_name" parent_obj_id
with
property value,
property value,
...
property value,
has attribute attribute ... attribute
;

The definition starts with the word Object and ends with a semicolon; in between are three major blocks of information:

  • immediately after the word Object is the header information;
  • the word with introduces the object’s properties;
  • the word has introduces the object’s attributes.

Object headers

An object header comprises up to three items, all optional:

  • An internal obj_id by which other objects refer to this object. It’s a single word (though it can contain digits and underscores) of up to thirty-two characters, and it must be unique within the game. You can omit the obj_id if this object isn’t referred to by any other objects.

    For example: bird, tree, top_of_tree.

  • An external_name, in double quotes, which is what the interpreter uses when referring to the object. It can be one or more words, and need not be unique (for instance, you might have several "Somewhere in the desert" rooms). Although not mandatory, it’s best to give every object an external_name. For example: "baby bird", "tall sycamore tree", "At the top of the tree".

  • The internal obj_id of another object which is the initial location of this object (its “parent” – see the next section) at the start of the game. This is omitted from objects which have no initial parent; it’s always omitted from a room.

    For example: the definition of the bird starts like this, specifying that at the start of the game, it can be found in the forest room (though later the player character will pick it up and move it around):

    Object   bird "baby bird" forest
    ...
    

    The tree starts like this; the only real difference is that, because the player character can’t move a scenery object, it’s always going to be in the clearing:

    Object   tree "tall sycamore tree" clearing
    ...
    

    Note

    There’s an alternative method for defining an object’s initial location, using “arrows” rather than the parent’s internal obj_id. For example, the definition of the bird could have started like this:

    Object   -> bird "baby bird"
    ...
    

    We don’t use the arrows method in this guide, though we do describe how it works in Setting up the object tree.

Object properties

An object’s property definitions are introduced by the with keyword. An object can have any number of properties, and they can be defined in any order. Each definition has two parts: a name, and a value; there’s a space between the two parts, and a comma at the end.

Think of each property as a variable which is specifically associated with that object. The variable’s initial setting is the supplied value; if necessary, it can be reset to other values during play (though in fact most property values don’t change in this way).

Here are examples of the properties that we’ve come across so far:

description "The nest is carefully woven of twigs and moss.",
e_to forest,
name 'baby' 'bird' 'nestling',
each_turn [; if (nest in branch) deadflag = 2; ],

By happy coincidence, those examples also demonstrate most of the different types of value which can be assigned to a property. The value associated with the description property in this particular example is a string of characters in double quotes; the value associated with this e_to property is the internal identity of an object; the name property is a bit unusual – its value is a list of dictionary words, each in single quotes; the each_turn property has a value which is an embedded routine (see Embedded routines). The only other type of value which is commonly found is a simple number; for example:

capacity 10,

In all, the library defines around forty-eight standard properties – like name and each_turn – which you can associate with your objects; there’s a complete list in Object properties. And in William Tell: in his prime we show you how to invent your own property variables.

Object attributes

An object’s attribute list is introduced by the has keyword. An object can have any number of attributes, and they can be listed in any order, with a space between each.

As with properties, you can think of each attribute as a variable which is specifically associated with that object. However, an attribute is a much more limited form of variable, since it can have only two possible states: present, and absent (also known as set/clear, on/off, or true/false; incidentally, a two-state variable like this is often called a flag). Initially, an attribute is either present (if you mention its name in the list) or absent (otherwise); if necessary, its state can change during play (and this is relatively common). We often say that a certain object currently has a certain attribute, or that conversely it hasn’t got it.

The attributes that we’ve come across so far are:

container light open scenery static supporter

Each of those answers a question: Is this object a container? Does it provide light? and so on. If the attribute is present then the answer is Yes; if the attribute isn’t present, the answer is No.

The library defines around thirty standard attributes, listed in Object attributes. Although you can devise additional attributes – see Common properties and attributes – in practice you seldom need to.

Object relationships – the object tree

Not only is your game composed entirely of objects, but also Inform takes great care to keep track of the relationships between those objects. By “relationship” we don’t mean that Walter is Wilhelm’s son, while Helga and Wilhelm are just good friends; it’s a much more comprehensive exercise in recording exactly where each object is located, relative to the other objects in the game.

Despite what we just said, Inform relationships are managed in terms of parent and child objects, though in a much broader sense than Wilhelm and Walter. When the player character is in a particular room – for example the forest – we can say that:

  • the forest object is the parent of the player object, or alternatively
  • the player object is a child of the forest object.

Also, if the player is carrying an object – for example the nest – we say that:

  • the player object is the parent of the nest object, or that
  • the nest object is a child of the player object.

Note the emphasis there: an object has exactly one parent (or no parent at all), but can have any number of child objects (including none).

For an example of an object having more than one child, think about the way we defined the nest and tree objects:

Object   nest "bird's nest" clearing
...

Object   tree "tall sycamore tree" clearing
...

We used the third of the header items to say that the clearing was the parent of the nest, and also that the clearing was the parent of the tree; that is, both nest and tree are child objects of the clearing.

Note

A “room” isn’t anything magical; it’s just an object which never has a parent, and which may from time to time have the player object as a child.

When we defined the bird, we placed it in the forest, like so:

Object   bird "baby bird" forest
...

We didn’t place any other objects in that room, so at the start of the game the forest was the parent of the bird (and the bird was the only child of the forest). But what happens when the player character, initially in the before_cottage room, goes EAST to the forest? Answer: the player’s parent is now the forest, and the forest has two children – the bird and the player. This is a key principle of the way Inform manages its objects: the parent–child relationships between objects change continuously, often dramatically, as the game progresses.

Another example of this: suppose the player character picks up the bird. This causes another change in the relationships. The bird is now a child of the player (and not of the forest), and the player is both a parent (of the bird) and a child (of the forest).

Here we show how the object relationships change during the course of the game. The straight lines represent parent–child relationships, with the parent object at the top of the line, and the child object at the bottom.

  1. At the start of the game:

    blockdiag before_ cottage forest clearing top_of_ tree player bird nest tree branch
  2. The player types: GO EAST

    blockdiag before_ cottage forest clearing top_of_ tree player bird nest tree branch
  3. The player types: TAKE THE BIRD

    blockdiag before_ cottage forest clearing top_of_ tree player bird nest tree branch
  4. The player types: GO NORTHEAST

    blockdiag before_ cottage forest clearing top_of_ tree player nest tree bird branch
  5. The player types: PUT BIRD IN NEST

    blockdiag before_ cottage forest clearing top_of_ tree player nest tree bird branch
  6. The player types: TAKE NEST

    blockdiag before_ cottage forest clearing top_of_ tree player tree bird nest branch
  7. The player types: UP

    blockdiag before_ cottage forest clearing top_of_ tree player branch bird nest tree
  8. The player types: PUT NEST ON BRANCH

    blockdiag before_ cottage forest clearing top_of_ tree player branch bird nest tree

In this short example, we’ve taken a lot of time and space to spell out exactly how the objects relationship patterns – generally known as the object tree – appear at each stage. Normally you wouldn’t bother with this much detail (a) because the interpreter does most of the work for you, and (b) because in a real game there are usually too many objects for you to keep track of. What’s important is that you understand the basic principles: at any moment in time an object either has no parent (which probably means either that it’s a room, or that it’s floating in hyperspace and not currently part of the game) or exactly one parent – the object that it’s “in” or “on” or “a part of”. However, there’s no restriction on the number of children that an object can have.

There’s a practical use for these relationships, covered in detail further on. As a designer, you can refer to the current parent or children of any given object with the parent, child and children routines, and this is one feature that you will be using frequently. There are also other routines associated with the object tree, to help you keep track of the objects or move them around. We’ll see them one by one in the next chapters. For a quick summary, see Objects.

Things in quotes

Inform makes careful distinction between double and single quotes.

Double quotes

Double quotes "..." surround a string – a letter, a word, a paragraph, or almost any number of characters – which you want the interpreter to display while the game is being played. You can use the tilde ~ to represent a double quote inside the string, and the circumflex ^ to represent a newline (line break) character. Upper-case and lower-case letters are treated as different.

A long string can be split over several lines; Inform transforms each line break (and any spaces around it) into a single space (extra spaces not at a line break are preserved, though). These two strings are equivalent:

"This is a      string of characters."

"This
  is
        a    string
                   of characters."

When the interpreter displays a long character string – for example, while describing a feature-packed room – it employs automatic word-wrapping to fit the text to the player’s screen. This is where you might insert ^ characters to force line breaks to appear, thus presenting the text as a series of paragraphs. So far, we’ve seen strings used as the value of a Constant:

Constant Headline
      "^A simple Inform example
       ^by Roger Firth and Sonja Kesserich.^";

which could equally have been defined thus:

Constant Headline
      "^A simple Inform example^by Roger Firth and Sonja Kesserich.^";

and as the value of an object description property:

description "Too young to fly, the nestling tweets helplessly.",

Later, you’ll find that they’re also very common in print statements.

Single quotes

Single quotes '...' surround a dictionary word. This has to be a single word – no spaces – and generally contains only letters (and occasionally numbers and hyphens), though you can use ^ to represent an apostrophe inside the word. Upper-case and lower-case letters are treated as identical; also, the interpreter normally looks only at the first nine characters of each word that the player types.

When the player types a command, the interpreter divides what was typed into individual words, which it then looks up in the dictionary. If it finds all the words, and they seem to represent a sensible course of action, that’s what happens next.

So far, we’ve seen dictionary words used as the values of an object name property:

name 'bird^s' 'nest' 'twigs' 'moss',

and indeed that’s just about the only place where they commonly occur. You’ll save yourself a lot of confusion by remembering the distinction: Double quotes for Output, Single quotes for Input (DOSI).

Routines and statements

A routine is a collection of statements, which are performed (or we often say “are executed”) at run-time by the interpreter. There are two types of routine, and about two dozen types of statement (there’s a complete list in Statements; see also Appendix E – Inform language).

Statements

A statement is an instruction telling the interpreter to perform a particular task – to “do something” – while the game is being played. A real game usually has lots and lots of statements, but so far we’ve encountered only a few. We saw:

location = before_cottage;

which is an example of an assignment statement, so-called because the equals sign = assigns a new value (the internal ID of our before_cottage room) to a variable (the global variable location which is part of the library). Later we saw:

if (nest in branch) deadflag = 2;

which is actually two statements: an assignment, preceded by an if statement:

if (nest in branch) ...

The if statement tests a particular condition; if the condition is true, the interpreter executes whatever statement comes next; if it isn’t true, the interpreter ignores the next statement. In this example, the interpreter is testing whether the nest object is “in” or “on” (which we now know means “is a child of”) the branch object. For most of the game, that condition is not true, and so the interpreter ignores the following statement. Eventually, when the condition becomes true, the interpreter executes that statement: it performs an assignment:

deadflag = 2;

which changes the value of the library variable deadflag from its current value to 2. Incidentally, if statements are often written on two lines, with the “controlled” statement indented. This makes it easier to read, but doesn’t change the way that it works:

if (nest in branch)
    deadflag = 2;

The thing that’s being controlled by the if statement doesn’t have to be an assignment; it can be any kind of statement. In fact, you can have lots of statements, not just one, controlled by an if statement. We’ll talk about these other possibilities later. For now, just remember that the only place where you’ll find statements are within standalone routines and embedded routines.

Standalone routines

A standalone routine is a series of statements, collected together and given a name. When the routine is “called” – by its given name – those statements are executed. Here’s the one that we’ve defined:

[ Initialise; location = before_cottage; ];

Because it’s such a tiny routine, we placed it all on a single line. Let’s rewrite it to use several lines (as with the if statement, this improves the readability, but doesn’t affect how it works):

[ Initialise;
    location = before_cottage;
];

The [ Initialise; is the start of the routine, and defines the name by which it can be “called”. The ]; is the end of the routine. In between are the statements – sometimes known as the body of the routine – which are executed when the routine is called. And how is that done? By a statement like this:

Initialise();

That single statement, the routine’s name followed by opening and closing parentheses, is all that it takes to call a routine. When it comes across a line like this, the interpreter executes the statements – in this example there’s only one, but there may be ten, twenty, even a hundred of them – in the body of the routine. Having done that, the interpreter resumes what it was doing, on the line following the Initialise(); call.

Note

You may have noticed that, although we’ve defined a routine named Initialise, we’ve never actually called it. Don’t worry – the routine is called, by the Inform library, right at the start of a game.

Embedded routines

An embedded routine is much like a standalone routine, though it doesn’t have a name and doesn’t end in a semicolon. This is the one that we defined:

[; if (nest in branch) deadflag = 2; ]

except that we didn’t write it in isolation like that: instead, we defined it to be the value of an object property:

each_turn [; if (nest in branch) deadflag = 2; ],

which would have worked just the same if we’d written it like this:

each_turn [;
    if (nest in branch)
        deadflag = 2;
],

All embedded routines are defined in this manner: as the value of an object property. That’s where they’re embedded – inside an object. The introductory characters [; maybe look a little odd, but it’s really only the same syntax as for a standalone routine, only without a name between the [ and ;.

For calling an embedded routine, thus causing the statements it contains to be executed, the method that we described for a standalone routine won’t work. An embedded routine has no name, and needs none; it’s automatically called by the library at appropriate moments, which are determined by the role of the property for which it is the value. In our example, that’s at the end of every turn in which the player character is in the same room as the branch. Later, we’ll see other examples of embedded routines, each designed to perform a task which is appropriate for the property whose value it is; we’ll also see that it is possible to call an embedded routine yourself, using an obj_id.property() syntax – in this example, we could call the routine by writing branch.each_turn(). There’s more about these topics in Routines and arguments, A diversion: working with routines and in Routines.

That ends our review of the ground covered in our first game. We’ll have more to say about most of this later, but we’re trying not to overload you with facts at this early stage. What we’d like you to do is to look back at the source of the game, and ensure that you can recognise all the elements which this chapter has described. Then, we’ll move on to fix a few of the game’s more important defects.