The phrases making up the basic Inform language, and in terms of which all other phrases and rules are defined.

• §4. Say phrases
• §14. Variables
• §17. Arithmetic
• §23. Control structures
• §32. Values
• §35. Text
• §40. Adaptive text
• §41. Data Structures
• §44. Lists
• §50. Functional Programming
• §52. Rulebooks and Activities
• §56. External Files
• §59. Use Options

§1. All phrases of the Inform language itself are defined here. The Standard Rules extension adds more, specialised for interactive fiction creation, but the bones of the language are here. For example, we will define what the plus sign will mean, and how Inform should compile "say N in words", where N is a number. Inform has no phrase definitions built in, as such; but it does contain assumptions about control structures such as "say ...", "repeat ...", "let ...", "otherwise ..." and so on will behave. Those, we would not be able to redefine in fundamentally different ways. In most respects, though, we are more or less free to define any language we like.

At first sight, these phrase definitions look little more than simple transliterations, and this was one source of early criticism of Inform 7. Phrases appeared to have very simplistic definitions, with the natural language simply being a verbose description of obviously equivalent C-like code. However, the simplicity is misleading, because the definitions below tend to conceal where the complexity of the translation process suddenly increases. If the preamble includes "(c - condition)", and the definition includes the expansion {c}, then the text forming c is translated in a way much more profound than any simple substitution process could describe. Type-checking also complicates the code produced below, since Inform automatically generates the code needed to perform run-time type checking at any point where doubt remains as to the phrase definition which must be used.

§2. Many of these phrases have what are called inline definitions, written using the (- and -) notation. Non-inline phrases are compiled as functions, which are then called when they need to be used. For example, "To say the magic number: say 17." compiles to a function, and when another phrase includes the instruction "say the magic number", that instruction compiles a call to this function. But an inline phrase instead generates Inter instructions to do something directly. (That something may in fact still be just a function call, but not always.) The Inter code to be generated is expressed by the contents of the definition between the (- and -) markers, and is written in a marked-up form of Inform 6 notation. This is much more concise and readable than if the Inter code were written out longhand, but it may give the misleading impression that Inform inline definitions can only produce Inform 6 code. That is not so: they produce Inter code, which can then be translated as needed.

Most of the definitions here also have annotations to positions in the main Inform documentation: for example, (documented at phs_s). This has no effect on the code compiled, and is used only when Inform generates certain problem messages; if the source text misuses the phrase, the problem can then give a reference to the relevant documentation. phs_s is a typical example of a "documentation token", and is only a label. See the source of the Inform documentation for how this markup is done.

§3. Unit tests for the phrases below have test case names beginning BIP-, which stands for "Basic Inform phrase". In fact, these come in pairs, one for each virtual machine we customarily generate code to. For example, BIP-Say tests the "say" phrase for the Z-machine target, and BIP-Say-G does the same for the Glulx target. But in the commentary below, test cases will be listed only once.

It follows that running intest -from inform7 BIP-%c+ will test all of these phrases on all platforms, since this regular expression matches all test case names beginning with "BIP-".

§4. Say phrases. We begin with saying phrases: the very first phrase to exist is the one printing a single value — literal text, a number, a time, an object, or really almost anything, since the vast majority of kinds in Inform are sayable. There used to be separate definitions for saying text, numbers and unicode characters here, but they were removed in June 2015 as being redundant. Though they did no harm, they made some problem messages longer than necessary by obliging them to cite a longer list of possible readings of a misread phrase.

The three inline definitions here neatly demonstrate the three sorts of things which appear inside (- and -). The definition for "To say s", which looks more familiar as the "[s]" text substitution, is straightforwardly Inform 6 notation. The definition for "To say (something - number) in words" is I6 notation except for the {something} part in braces: this expands to the value used by the code causing compilation. For example, if the code to be compiled is "say 17 in words" then {something} here would expand to the constant 17. The definition for "To say (val)" is much more complex than I6 notation could convey, and so a more complex escape notation is needed, {-say:val:K}, which tells Inform to compile code which will say val with whatever method is appropriate to its kind K. For documentation on these escape notations, see the core Inform source code.

The global variable say__n tracks the last number printed. For the "in words" definition, we need to set it by hand, since Inform doesn't otherwise realise that number-printing is what we are doing here. For definitions of functions such as STextSubstitution, see the source for BasicInformKit, which is also where say__n is defined.

See test case BIP-Say.

Part Three - Phrasebook

Chapter 1 - Saying

Section 1 - Saying Values

To say (val - sayable value of kind K)
    (documented at ph_say):
    (- {-say:val:K} -).
To say (something - number) in words
    (documented at phs_numwords):
    (- print (number) say__n=({something}); -).
To say s
    (documented at phs_s):
    (- STextSubstitution(); -).

§5. "Showme" is a debugging version of "say" which can print some version of the value, and the kind, of just about anything.

See test case BIP-Showme.

To showme (val - value)
    (documented at ph_showme):
    (- {-show-me:val} -).

§6. Now some fun with number bases:

To say (N - a number) in hexadecimal
    (documented at phs_inbase):
    (- PrintInBase({N}, 16); -).

To say (N - a number) in decimal
    (documented at phs_inbase):
    (- PrintInBase({N}, 10); -).

To say (N - a number) in octal
    (documented at phs_inbase):
    (- PrintInBase({N}, 8); -).

To say (N - a number) in binary
    (documented at phs_inbase):
    (- PrintInBase({N}, 2); -).

To say (N - a number) in base (B - a number)
    (documented at phs_inbase):
    (- PrintInBase({N}, {B}); -).

To say (N - a number) in (M - a number) digit/digits
    (documented at phs_indigits):
    (- PrintInBase({N}, 10, {M}); -).

To say (N - a number) in (M - a number) hexadecimal digit/digits
    (documented at phs_inbaseindigits):
    (- PrintInBase({N}, 16, {M}); -).

To say (N - a number) in (M - a number) decimal digit/digits
    (documented at phs_inbaseindigits):
    (- PrintInBase({N}, 10, {M}); -).

To say (N - a number) in (M - a number) octal digit/digits
    (documented at phs_inbaseindigits):
    (- PrintInBase({N}, 8, {M}); -).

To say (N - a number) in (M - a number) binary digit/digits
    (documented at phs_inbaseindigits):
    (- PrintInBase({N}, 2, {M}); -).

To say (N - a number) in (M - a number) base (B - a number) digit/digits
    (documented at phs_inbaseindigits):
    (- PrintInBase({N}, {B}, {M}); -).

To say (N - a number) in unsigned decimal
    (documented at phs_inunsigneddecimal):
    (- PrintInBase({N}, 10, 1); -).

To say (N - a number) in (M - a number) unsigned decimal digit/digits
    (documented at phs_inunsigneddecimaldigits):
    (- PrintInBase({N}, 10, {M}); -).

§7. Objects are the most difficult things to say, because of the elaborate apparatus for managing their natural-language representations. In particular, we need to say them with a definite or indefinite article, which can either be capitalised or not, and as part of that we need to keep track of whether they are proper nouns; in languages other than English, there are also gender and case to worry about.

Note that "To say ..." phrases are case sensitive on the first word, so that "to say a something" and "to say A something" are different.

A curiosity of Inform 6's syntax, arising I think mostly from the need to save property memory in "Curses" (1993), the work of IF for which Inform 1 had been created, is that it lacks a print (A) ... statement. The omission is made good by using a routine in the template library instead.

See test case BIP-SayName.

Section 2 - Saying Names

To say a (something - object)
    (documented at phs_a):
    (- print (a) {something}; -).
To say an (something - object)
    (documented at phs_a):
    (- print (a) {something}; -).
To say A (something - object)
    (documented at phs_A):
    (- CIndefArt({something}); -).
To say An (something - object)
    (documented at phs_A):
    (- CIndefArt({something}); -).
To say the (something - object)
    (documented at phs_the):
    (- print (the) {something}; -).
To say The (something - object)
    (documented at phs_The):
    (- print (The) {something}; -).

§8. Now some text substitutions which are the equivalent of escape characters. (In double-quoted I6 text, the notation for a literal quotation mark is a tilde ~.) Note the use of the "-- running on" annotation, which tells Inform that a text substitution should not cause a new-line.

See test case BIP-SaySpecial.

Section 3 - Saying Special Characters

To say bracket -- running on
    (documented at phs_bracket):
    (- print "["; -).
To say close bracket -- running on
    (documented at phs_closebracket):
    (- print "]"; -).
To say apostrophe/' -- running on
    (documented at phs_apostrophe):
    (- print "'"; -).
To say quotation mark -- running on
    (documented at phs_quotemark):
    (- print "~"; -).

§9. For an explanation of the paragraph breaking algorithm, see the template file "Printing.i6t".

See test case BIP-SayParagraphing.

Section 4 - Saying Line and Paragraph Breaks

To say line break -- running on
    (documented at phs_linebreak):
    (- new_line; -).
To say no line break -- running on
    (documented at phs_nolinebreak):
    do nothing.
To say conditional paragraph break -- running on
    (documented at phs_condparabreak):
    (- DivideParagraphPoint(); -).
To say paragraph break -- running on
    (documented at phs_parabreak):
    (- DivideParagraphPoint(); new_line; -).
To say run paragraph on -- running on
    (documented at phs_runparaon):
    (- RunParagraphOn(); -).
To decide if a paragraph break is pending
    (documented at ph_breakpending):
    (- (say__p) -).

§10. Now for "[if ...]", which expands into a rather assembly-language-like usage of jump statements, I6's form of goto. For instance, the text "[if the score is 10]It's ten![otherwise]It's not ten, alas." compiles thus:

    if (~~(score == 10)) jump L_Say3;
        ...
    jump L_SayX2; .L_Say3;
        ...
    .L_Say4; .L_SayX2;

Though labels actually have local namespaces in I6 routines, we use globally unique labels throughout the whole program: compiling the same phrase again would involve say labels 5 and 6 and "say exit" label 3. This example text demonstrates the reason we jump about, rather than making use of if... else... and bracing groups of statements: it is legal in I7 either to conclude with or to omit the "[end if]". (If statements in I6 compile to jump instructions in any event, and on our virtual machines there is no speed penalty for branches.) We also need the same definitions to accommodate what amounts to a switch statement. The trickier text "[if the score is 10]It's ten![otherwise if the score is 8]It's eight?[otherwise]It's not ten, alas." comes out as:

    if (~~(score == 10)) jump L_Say5;
        ...
    jump L_SayX3; .L_Say5; if (~~(score == 8)) jump L_Say6;
        ...
    jump L_SayX3; .L_Say6;
        ...
    .L_Say7; .L_SayX3;

In either form of the construct, control passes into at most one of the pieces of text. The terminal labels (the two on the final line) are automatically generated; often — when there is a simple "otherwise" or "end if" to conclude the construct — they are not needed, but labels are quick to process in I6, are soon discarded from I6's memory when not needed any more, and compile no code.

We assume in each case that the next say label number to be free is always the start of the next block, and that the next say exit label number is always the one at the end of the current construct. This is true because Inform does not allow "say if" to be nested.

See test case BIP-SayIf.

Section 5 - Saying If and Otherwise

To say if (c - condition)
    (documented at phs_if): (-
    if (~~({c})) jump {-label:Say};
        -).
To say unless (c - condition)
    (documented at phs_unless): (-
    if ({c}) jump {-label:Say};
        -).
To say otherwise/else if (c - condition)
    (documented at phs_elseif): (-
    jump {-label:SayX}; .{-label:Say}{-counter-up:Say}; if (~~({c})) jump {-label:Say};
        -).
To say otherwise/else unless (c - condition)
    (documented at phs_elseunless): (-
    jump {-label:SayX}; .{-label:Say}{-counter-up:Say}; if ({c}) jump {-label:Say};
        -).
To say otherwise
    (documented at phs_otherwise): (-
    jump {-label:SayX}; .{-label:Say}{-counter-up:Say};
        -).
To say else
    (documented at phs_otherwise): (-
    jump {-label:SayX}; .{-label:Say}{-counter-up:Say};
        -).
To say end if
    (documented at phs_endif): (-
    .{-label:Say}{-counter-up:Say}; .{-label:SayX}{-counter-up:SayX};
        -).
To say end unless
    (documented at phs_endunless): (-
    .{-label:Say}{-counter-up:Say}; .{-label:SayX}{-counter-up:SayX};
        -).

§11. The other control structure: the random variations form of saying. This part of the Inform design was in effect contributed by the community: it reimplements a form of Jon Ingold's former extension Text Variations, which itself built on code going back to the days of I6.

The head phrase has one of the most complicated definitions suppled with Inform, but is actually documented fairly explicitly in the Extensions chapter of "Writing with Inform", so we won't repeat all that here. Essentially it uses its own allocated cell of storage in an array to remember a state between uses, and compiles as a switch statement based on the current state.

See test case BIP-SayOneOf.

Section 6 - Saying one of

To say one of -- beginning say_one_of (documented at phs_oneof): (-
    {-counter-makes-array:say_one_of}
    {-counter-makes-array:say_one_flag}
    if ({-counter-storage:say_one_flag}-->{-counter:say_one_flag} == false) {
        {-counter-storage:say_one_of}-->{-counter:say_one_of} = {-final-segment-marker}({-counter-storage:say_one_of}-->{-counter:say_one_of}, {-segment-count});
        {-counter-storage:say_one_flag}-->{-counter:say_one_flag} = true;
    }
    if (say__comp == false) {-counter-storage:say_one_flag}-->{-counter:say_one_flag}{-counter-up:say_one_flag} = false;
    switch (({-counter-storage:say_one_of}-->{-counter:say_one_of}{-counter-up:say_one_of})%({-segment-count}+1)-1)
{-open-brace}
        0: -).
To say or -- continuing say_one_of (documented at phs_or):
    (- {-segment-count}: -).
To say at random -- ending say_one_of with marker I7_SOO_RAN (documented at phs_random):
    (- {-close-brace} -).
To say purely at random -- ending say_one_of with marker I7_SOO_PAR (documented at phs_purelyrandom):
    (- {-close-brace} -).
To say then at random -- ending say_one_of with marker I7_SOO_TRAN (documented at phs_thenrandom):
    (- {-close-brace} -).
To say then purely at random -- ending say_one_of with marker I7_SOO_TPAR (documented at phs_thenpurelyrandom):
    (- {-close-brace} -).
To say sticky random -- ending say_one_of with marker I7_SOO_STI (documented at phs_sticky):
    (- {-close-brace} -).
To say as decreasingly likely outcomes -- ending say_one_of with marker I7_SOO_TAP (documented at phs_decreasing):
    (- {-close-brace} -).
To say in random order -- ending say_one_of with marker I7_SOO_SHU (documented at phs_order):
    (- {-close-brace} -).
To say cycling -- ending say_one_of with marker I7_SOO_CYC (documented at phs_cycling):
    (- {-close-brace} -).
To say stopping -- ending say_one_of with marker I7_SOO_STOP (documented at phs_stopping):
    (- {-close-brace} -).

To say first time -- beginning say_first_time (documented at phs_firsttime):
    (- {-counter-makes-array:say_first_time}
    if ((say__comp == false) && (({-counter-storage:say_first_time}-->{-counter:say_first_time}{-counter-up:say_first_time})++ == 0)) {-open-brace}
        -).
To say only -- ending say_first_time (documented at phs_firsttime):
    (- {-close-brace} -).

§12. Now some visual effects, which may or may not be rendered the way the user hopes: that's partly up to the virtual machine, unfortunately.

See test case BIP-SayFonts, though since intest runs on plain text only, you may need to run this in the Inform application to be convinced.

Section 7 - Saying Fonts and Visual Effects

To say bold type -- running on
    (documented at phs_bold):
    (- style bold; -).
To say italic type -- running on
    (documented at phs_italic):
    (- style underline; -).
To say roman type -- running on
    (documented at phs_roman):
    (- style roman; -).
To say fixed letter spacing -- running on
    (documented at phs_fixedspacing):
    (- font off; -).
To say variable letter spacing -- running on
    (documented at phs_varspacing):
    (- font on; -).

§13. These are lists in the sense of the "list of" kind of value constructor, and the first two phrases here might list any values, not just objects.

See test case BIP-SayLists.

Section 8 - Saying Lists of Values

To say (L - a list of values) in brace notation
    (documented at phs_listbraced):
    (- LIST_OF_TY_Say({-by-reference:L}, 1); -).
To say (L - a list of objects) with definite articles
    (documented at phs_listdef):
    (- LIST_OF_TY_Say({-by-reference:L}, 2); -).
To say (L - a list of objects) with indefinite articles
    (documented at phs_listindef):
    (- LIST_OF_TY_Say({-by-reference:L}, 3); -).

§14. Variables. The "now" phrase can do an extraordinary range of things, and is more or less a genie granting one wish.

The "whether or not" phrase exists only to convert a condition to a value, and is needed because I7 does not silently cast from one to the other in the way that C would.

See test case BIP-Now.

Chapter 2 - Conditions and Variables

Section 1 - Conditions

To now (cn - condition)
    (documented at ph_now):
    (- {cn} -).
To decide what truth state is whether or not (C - condition)
    (documented at ph_whether):
    (- ({C}) -).

§15. Assignment is probably the most difficult thing the type-checker has to cope with, since "let" has to work when applied to both unknown names (it creates a new variable) and existing ones (kind of value permitting). There are also four different ways to create with "let", and two to use existing variables. Note that the "given by" forms are not strictly speaking assignments at all; the value placed in t is found by solving the equation Q. This does require special typechecking, but of a different kind to that requested by "(assignment operation)". All of which makes the "To let" section here only slightly shorter than John Galsworthy's Forsyte novel of the same name.

See test case BIP-Let.

Section 2 - Assigning Temporary Variables

To let (t - nonexisting variable) be (u - value)
    (assignment operation)
    (documented at ph_let): (-
        {-unprotect:t}
        {-copy:t:u}
    -).
To let (t - nonexisting variable) be (u - name of kind of value)
    (assignment operation)
    (documented at ph_letdefault): (-
        {-unprotect:t}
        {-initialise:t}
    -).
To let (t - nonexisting variable) be (u - description of relations of values
    of kind K to values of kind L)
    (assignment operation)
    (documented at ph_letrelation): (-
        {-unprotect:t}
        {-initialise:t}
        {-now-matches-description:t:u};
    -).
To let (t - nonexisting variable) be given by (Q - equation name)
    (documented at ph_letequation): (-
        {-unprotect:t}
        {-primitive-definition:solve-equation};
    -).

To let (t - existing variable) be (u - value)
    (assignment operation)
    (documented at ph_let): (-
        {-copy:t:u}
    -).
To let (t - existing variable) be given by (Q - equation name)
    (documented at ph_letequation): (-
        {-primitive-definition:solve-equation};
    -).

§16. It is not explicit in the following definitions that Inform should typecheck that the values held by these storage objects can be incremented or decremented (as an object, say, cannot, but a number can): Inform nevertheless contains code which does this.

See test case BIP-Increase.

Section 3 - Increase and Decrease

To increase (S - storage) by (w - value)
    (assignment operation)
    (documented at ph_increase): (-
        {-copy:S:+w};
    -).
To decrease (S - storage) by (w - value)
    (assignment operation)
    (documented at ph_decrease): (-
        {-copy:S:-w};
    -).
To increment (S - storage)
    (documented at ph_increment): (-
        {-copy:S:+};
    -).
To decrement (S - storage)
    (documented at ph_decrement): (-
        {-copy:S:-};
    -).

§17. Arithmetic. There are nine arithmetic operations, internally numbered 0 upwards, and given verbal forms below. These are handled unusually in the type-checker because they need to be more polymorphic than most phrases: Inform uses dimension-checking to determine the kind of value resulting. (Thus a height times a number is another height, and so on.)

The totalling code (12) is not structly to do with arithmetic in the same way, but it's needed to flag the phrase for the Inform typechecker's special attention.

See test case BIP-ArithmeticOperations.

Chapter 2 - Arithmetic

Section 1 - Arithmetic Operations

To decide which arithmetic value is (X - arithmetic value) + (Y - arithmetic value)
    (arithmetic operation 0)
    (documented at ph_plus):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) plus (Y - arithmetic value)
    (arithmetic operation 0)
    (documented at ph_plus):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) - (Y - arithmetic value)
    (arithmetic operation 1)
    (documented at ph_minus):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) minus (Y - arithmetic value)
    (arithmetic operation 1)
    (documented at ph_minus):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) * (Y - arithmetic value)
    (arithmetic operation 2)
    (documented at ph_times):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) times (Y - arithmetic value)
    (arithmetic operation 2)
    (documented at ph_times):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) multiplied by (Y - arithmetic value)
    (arithmetic operation 2)
    (documented at ph_times):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) / (Y - arithmetic value)
    (arithmetic operation 3)
    (documented at ph_divide):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) divided by (Y - arithmetic value)
    (arithmetic operation 3)
    (documented at ph_divide):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is remainder after dividing (X - arithmetic value)
    by (Y - arithmetic value)
    (arithmetic operation 4)
    (documented at ph_remainder):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is (X - arithmetic value) to the nearest (Y - arithmetic value)
    (arithmetic operation 5)
    (documented at ph_nearest):
    (- ({-arithmetic-operation:X:Y}) -).
To decide which arithmetic value is the square root of (X - arithmetic value)
    (arithmetic operation 6)
    (documented at ph_squareroot):
    (- ({-arithmetic-operation:X}) -).
To decide which arithmetic value is the cube root of (X - arithmetic value)
    (arithmetic operation 8)
    (documented at ph_cuberoot):
    (- ({-arithmetic-operation:X}) -).
To decide which arithmetic value is total (p - arithmetic value valued property)
    of (S - description of values)
    (arithmetic operation 12)
    (documented at ph_total):
    (- {-primitive-definition:total-of} -).

§18. Real numbers are not available in the Z-machine, but they are likely to be available everywhere else, i.e., on any other platform Inform may target in future.

See test case BIP-SayRealNumbers-G, which has no Z-machine counterpart.

Section 2 - Saying Real Numbers (not for Z-machine)

To say (R - a real number) to (N - number) decimal places
    (documented at phs_realplaces):
    (- Float({R}, {N}); -).
To say (R - a real number) in decimal notation
    (documented at phs_decimal):
    (- FloatDec({R}); -).
To say (R - a real number) to (N - number) decimal places in decimal notation
    (documented at phs_decimalplaces):
    (- FloatDec({R}, {N}); -).
To say (R - a real number) in scientific notation
    (documented at phs_scientific):
    (- FloatExp({R}); -).
To say (R - a real number) to (N - number) decimal places in scientific notation
    (documented at phs_scientificplaces):
    (- FloatExp({R}, {N}); -).

§19. A number of miscellaneous mathematical functions follow, for real numbers only; these are tested as part of BIP-ArithmeticOperations-G, already mentioned above. Note that we do not need to define real versions of addition, multiplication and so on: the above definitions are polymorphic enough to have done that already.

Section 3 - Real Arithmetic (not for Z-machine)

To decide which real number is the reciprocal of (R - a real number)
    (documented at ph_reciprocal):
    (- REAL_NUMBER_TY_Reciprocal({R}) -).
To decide which real number is the absolute value of (R - a real number)
    (documented at ph_absolutevalue)
    (this is the abs function):
    (- REAL_NUMBER_TY_Abs({R}) -).
To decide which real number is the real square root of (R - a real number)
    (arithmetic operation 7)
    (documented at ph_realsquareroot)
    (this is the root function inverse to rsqr):
    (- REAL_NUMBER_TY_Root({R}) -).
To decide which real number is the real square of (R - a real number)
    (this is the rsqr function inverse to root):
    let x be given by x = R^2 where x is a real number;
    decide on x.
To decide which real number is the ceiling of (R - a real number)
    (documented at ph_ceiling)
    (this is the ceiling function):
    (- REAL_NUMBER_TY_Ceiling({R}) -).
To decide which real number is the floor of (R - a real number)
    (documented at ph_floor)
    (this is the floor function):
    (- REAL_NUMBER_TY_Floor({R}) -).
To decide which number is (R - a real number) to the nearest whole number
    (documented at ph_nearestwholenumber)
    (this is the int function):
    (- REAL_NUMBER_TY_to_NUMBER_TY({R}) -).

§20. And these are tested in BIP-Exponentials-G.

Section 4 - Exponential Functions (not for Z-machine)

To decide which real number is the natural/-- logarithm of (R - a real number)
    (documented at ph_logarithm)
    (this is the log function inverse to exp):
    (- REAL_NUMBER_TY_Log({R}) -).
To decide which real number is the logarithm to base (N - a number) of (R - a real number)
    (documented at ph_logarithmto):
    (- REAL_NUMBER_TY_BLog({R}, {N}) -).
To decide which real number is the exponential of (R - a real number)
    (documented at ph_exp)
    (this is the exp function inverse to log):
    (- REAL_NUMBER_TY_Exp({R}) -).
To decide which real number is (R - a real number) to the power (P - a real number)
    (documented at ph_power):
    (- REAL_NUMBER_TY_Pow({R}, {P}) -).

§21. And these are tested in BIP-Trigonometry-G.

Section 5 - Trigonometric Functions (not for Z-machine)

To decide which real number is (R - a real number) degrees
    (documented at ph_degrees):
    (- REAL_NUMBER_TY_Times({R}, $+0.0174532925) -).

To decide which real number is the sine of (R - a real number)
    (documented at ph_sine)
    (this is the sin function inverse to arcsin):
    (- REAL_NUMBER_TY_Sin({R}) -).
To decide which real number is the cosine of (R - a real number)
    (documented at ph_cosine)
    (this is the cos function inverse to arccos):
    (- REAL_NUMBER_TY_Cos({R}) -).
To decide which real number is the tangent of (R - a real number)
    (documented at ph_tangent)
    (this is the tan function inverse to arctan):
    (- REAL_NUMBER_TY_Tan({R}) -).
To decide which real number is the arcsine of (R - a real number)
    (documented at ph_arcsine)
    (this is the arcsin function inverse to sin):
    (- REAL_NUMBER_TY_Arcsin({R}) -).
To decide which real number is the arccosine of (R - a real number)
    (documented at ph_arccosine)
    (this is the arccos function inverse to cos):
    (- REAL_NUMBER_TY_Arccos({R}) -).
To decide which real number is the arctangent of (R - a real number)
    (documented at ph_arctangent)
    (this is the arctan function inverse to tan):
    (- REAL_NUMBER_TY_Arctan({R}) -).

§22. And these are tested in BIP-Hyperbolics-G.

Section 6 - Trigonometric Functions (not for Z-machine)

To decide which real number is the hyperbolic sine of (R - a real number)
    (documented at ph_hyperbolicsine)
    (this is the sinh function inverse to arcsinh):
    (- REAL_NUMBER_TY_Sinh({R}) -).
To decide which real number is the hyperbolic cosine of (R - a real number)
    (documented at ph_hyperboliccosine)
    (this is the cosh function inverse to arccosh):
    (- REAL_NUMBER_TY_Cosh({R}) -).
To decide which real number is the hyperbolic tangent of (R - a real number)
    (documented at ph_hyperbolictangent)
    (this is the tanh function inverse to arctanh):
    (- REAL_NUMBER_TY_Tanh({R}) -).
To decide which real number is the hyperbolic arcsine of (R - a real number)
    (documented at ph_hyperbolicarcsine)
    (this is the arcsinh function inverse to sinh):
    let x be given by x = log(R + root(R^2 + 1)) where x is a real number;
    decide on x.
To decide which real number is the hyperbolic arccosine of (R - a real number)
    (documented at ph_hyperbolicarccosine)
    (this is the arccosh function inverse to cosh):
    let x be given by x = log(R + root(R^2 - 1)) where x is a real number;
    decide on x.
To decide which real number is the hyperbolic arctangent of (R - a real number)
    (documented at ph_hyperbolicarctangent)
    (this is the arctanh function inverse to tanh):
    let x be given by x = 0.5*(log(1+R) - log(1-R)) where x is a real number;
    decide on x.

§23. Control structures. The term "control structure" conjures up the thought of conditionals and loops, and we'll get to those, but we'll begin with the equivalent of the C language's return statement: ending a function call with some value as an outcome. Inform calls this "deciding" something, since in Inform programs functions returning values are usually quite functional: that is, their point is what value they return, rather than the side-effects of what they did.

Note that returning a value has to invoke the type-checker to ensure that the return value matches the kind of value expected. This certainly rejects the phrase if it's used in a definition which isn't meant to be deciding a value at all, so an "in... only" clause is not needed.

The IF-form of Inform allows the antique syntaxes "yes" and "no" as synonyms for "decide yes" and "decide no"; these are not present in Basic Inform, and are defined in the Standard Rules (and only to keep old source text working).

See test case BIP-Decide.

Chapter 3 - Control

Section 1 - Deciding Outcomes

To decide yes
    (documented at ph_yes):
    (- rtrue; -) - in to decide if only.
To decide no
    (documented at ph_no):
    (- rfalse; -) - in to decide if only.

To stop (documented at ph_stop):
    (- rtrue; -) - in to only.

To decide on (something - value)
    (documented at ph_decideon):
    (- return {-return-value:something}; -).

§24. While "unless" is supposed to be exactly like "if" but with the reversed sense of the condition, that isn't quite true. For example, there is no "unless ... then ...": logical it might be, English it is not.

The switch form of "if" is subtly different, and here again "unless" is not allowed in its place.

As with some other control structures, the definitions here are somewhat partial, and made up for by direct code in the compiler. (There's a limit to how much a general syntax for phrases can encode control phrases.)

See test case BIP-If.

Section 2 - If and Unless

To if (c - condition) begin -- end conditional
    (documented at ph_if):
    (- {c}  -).
To unless (c - condition) begin -- end conditional
    (documented at ph_unless):
    (- (~~{c})  -).
To if (V - value) is begin -- end conditional
    (documented at ph_switch):
    (-  -).

§25. "Do nothing" is a curious feature for a high-level programming language (C, for example, does not have a NOP function); it entered Inform in the earliest days, when it was useful mainly when natural language syntax had painted users into a corner. (In the examples, it used to be used when conditions were awkward to negate — if condition, do nothing, otherwise blah blah blah — but the creation of "unless" made it possible to remove most of the "do nothing"s.) It is now hardly ever useful.

To do nothing (documented at ph_nothing):
    (- ; -).

§26. After all that, the while loop is simplicity itself. Perhaps the presence of "unless" for "if" argues for a similarly negated form, "until" for "while", but users haven't yet petitioned for this.

See test case BIP-Loops.

Section 3 - While and Repeat

To while (c - condition) begin -- end loop
    (documented at ph_while):
    (- while {c}  -).

§27. The repeat loop looks like a single construction, but isn't, because the range can be given in four fundamentally different ways (and the loop variable then has a different kind of value accordingly). First, the equivalents of BASIC's for loop and of Inform 6's objectloop, respectively:

To repeat with (loopvar - nonexisting K variable)
    running from (v - arithmetic value of kind K) to (w - K) begin -- end loop
    (documented at ph_repeat):
        (- for ({loopvar}={v}: {loopvar}<={w}: {loopvar}++)  -).
To repeat with (loopvar - nonexisting K variable)
    running from (v - enumerated value of kind K) to (w - K) begin -- end loop
    (documented at ph_repeat):
        (- for ({loopvar}={v}: {loopvar}<={w}: {loopvar}={-next-routine:K}({loopvar}))  -).
To repeat with (loopvar - nonexisting K variable)
    running through (OS - description of values of kind K) begin -- end loop
    (documented at ph_runthrough):
        (- {-primitive-definition:repeat-through} -).
To repeat with (loopvar - nonexisting object variable)
    running through (L - list of values) begin -- end loop
    (documented at ph_repeatlist):
        (- {-primitive-definition:repeat-through-list} -).

§28. The following are all repeats where the range is the set of rows of a table, taken in some order, and the repeat variable — though it does exist — is never specified since the relevant row is instead the one selected during each iteration of the loop.

To repeat through (T - table name) begin -- end loop
    (documented at ph_repeattable): (-
        @push {-my:ct_0}; @push {-my:ct_1};
        for ({-my:1}={T}, {-my:2}=1, ct_0={-my:1}, ct_1={-my:2}:
            {-my:2}<=TableRows({-my:1}):
            {-my:2}++, ct_0={-my:1}, ct_1={-my:2})
            if (TableRowIsBlank(ct_0, ct_1)==false)
                {-block}
        @pull {-my:ct_1}; @pull {-my:ct_0};
    -).
To repeat through (T - table name) in reverse order begin -- end loop
    (documented at ph_repeattablereverse): (-
        @push {-my:ct_0}; @push {-my:ct_1};
        for ({-my:1}={T}, {-my:2}=TableRows({-my:1}), ct_0={-my:1}, ct_1={-my:2}:
            {-my:2}>=1:
            {-my:2}--, ct_0={-my:1}, ct_1={-my:2})
            if (TableRowIsBlank(ct_0, ct_1)==false)
                {-block}
        @pull {-my:ct_1}; @pull {-my:ct_0};
    -).
To repeat through (T - table name) in (TC - table column) order begin -- end loop
    (documented at ph_repeattablecol): (-
        @push {-my:ct_0}; @push {-my:ct_1};
        for ({-my:1}={T}, {-my:2}=TableNextRow({-my:1}, {TC}, 0, 1), ct_0={-my:1}, ct_1={-my:2}:
            {-my:2}~=0:
            {-my:2}=TableNextRow({-my:1}, {TC}, {-my:2}, 1), ct_0={-my:1}, ct_1={-my:2})
                {-block}
        @pull {-my:ct_1}; @pull {-my:ct_0};
    -).
To repeat through (T - table name) in reverse (TC - table column) order begin -- end loop
    (documented at ph_repeattablecolreverse): (-
        @push {-my:ct_0}; @push {-my:ct_1};
        for ({-my:1}={T}, {-my:2}=TableNextRow({-my:1}, {TC}, 0, -1), ct_0={-my:1}, ct_1={-my:2}:
            {-my:2}~=0:
            {-my:2}=TableNextRow({-my:1}, {TC}, {-my:2}, -1), ct_0={-my:1}, ct_1={-my:2})
                {-block}
        @pull {-my:ct_1}; @pull {-my:ct_0};
    -).

§29. And this loops through the lines of a text file stored internally.

To repeat with (loopvar - nonexisting text variable)
    running through (F - internal file) begin -- end loop:
    (-
        for ({-my:1} = InternalFileIO_Line({-by-reference:loopvar}, {F}): {-my:1}:
            {-my:1} = InternalFileIO_Line({-by-reference:loopvar}, {F}))
            {-block}
    -).

§30. The equivalent of break or continue in C or I6, or of last or next in Perl. Here "in loop" means "in any of the forms of while or repeat".

See test case BIP-Break.

Section 4 - Loop Flow

To break -- in loop
    (documented at ph_break):
    (- {-primitive-definition:break} -).
To next -- in loop
    (documented at ph_next):
    (- continue; -).

§31. The following innocent-looking definition throws a problem message if the RTP label is not (a) literal text and (b) the leafname of a Markdown file (once .md has been added) in the RTPs subdirectory of the current extension directory: it can only be used from with a directory-format extension.

Section 5 - Run-Time Problems

To issue the run-time problem (pcode - text):
    (- IssueRTP({-rtp-code: pcode}, -1, {-rtp-location: pcode}); -).

§32. Values. Some of the things we can do with enumerations, others being listed under randomness below.

See test case BIP-Enumerations.

Chapter 4 - Values

Section 1 - Enumerations

To decide which number is number of (S - description of values)
    (documented at ph_numberof):
    (- {-primitive-definition:number-of} -).
To decide what number is the numerical value of (X - enumerated value)
    (documented at ph_numericalvalue):
    (- {X} -).
To decide what number is the sequence number of (X - enumerated value of kind K)
    (documented at ph_sequencenumber):
    (- {-indexing-routine:K}({X}) -).
To decide which K is (name of kind of enumerated value K) after (X - K)
    (documented at ph_enumafter):
    (- {-next-routine:K}({X}) -).
To decide which K is (name of kind of enumerated value K) before (X - K)
    (documented at ph_enumbefore):
    (- {-previous-routine:K}({X}) -).
To decide which K is the first value of (name of kind of enumerated value K)
    (documented at ph_enumfirst):
    decide on the default value of K.
To decide which K is the last value of (name of kind of enumerated value K)
    (documented at ph_enumlast):
    decide on K before the default value of K.

§33. Random numbers and random items chosen from sets of objects matching a given description ("a random closed door").

See test case BIP-Randomness.

Section 2 - Randomness

To decide which K is a/-- random (S - description of values of kind K)
    (documented at ph_randomdesc):
    (- {-primitive-definition:random-of} -).
To decide which K is a random (name of kind of arithmetic value K) between (first value - K) and (second value - K)
    (documented at ph_randombetween):
    (- {-ranger-routine:K}({first value}, {second value}) -).
To decide which K is a random (name of kind of arithmetic value K) from (first value - K) to (second value - K)
    (documented at ph_randombetween):
    (- {-ranger-routine:K}({first value}, {second value}) -).
To decide which K is a random (name of kind of enumerated value K) between (first value - K) and (second value - K)
    (documented at ph_randombetween):
    (- {-ranger-routine:K}({first value}, {second value}) -).
To decide which K is a random (name of kind of enumerated value K) from (first value - K) to (second value - K)
    (documented at ph_randombetween):
    (- {-ranger-routine:K}({first value}, {second value}) -).
To decide whether a random chance of (N - number) in (M - number) succeeds
    (documented at ph_randomchance):
    (- (GenerateRandomNumber(1, {M}) <= {N}) -).
To seed the random-number generator with (N - number)
    (documented at ph_seed):
    (- VM_Seed_RNG({N}); -).

§34. A novel feature of Inform is that there is a default value of any kind: for example, it is 0 for a number, or the empty text for text. When Inform compiles a value of a given kind but isn't told what value to compile, it always chooses the default, which is why the following definition works.

See test case BIP-DefaultValues.

Section 3 - Default Values

To decide what K is the default value of (V - name of kind of value of kind K)
    (documented at ph_defaultvalue):
    (- {-new:K} -).

§35. Text. Inform programs swim in a sea of texts, and most of the ways to make text involve substitutions; so phrases to manipulate text are nice to have, but are bonuses rather than being the essentials.

As repetitive as the following is, it's much simpler and less prone to possible namespace trouble if we don't define kinds of value for the different structural levels of text (character, word, punctuated word, etc.).

See test case BIP-Texts.

Chapter 5 - Text

Section 1 - Breaking down text

To decide what number is the number of characters in (T - text)
    (documented at ph_numchars):
    (- TEXT_TY_BlobAccess({-by-reference:T}, CHR_BLOB) -).
To decide what number is the number of words in (T - text)
    (documented at ph_numwords):
    (- TEXT_TY_BlobAccess({-by-reference:T}, WORD_BLOB) -).
To decide what number is the number of punctuated words in (T - text)
    (documented at ph_numpwords):
    (- TEXT_TY_BlobAccess({-by-reference:T}, PWORD_BLOB) -).
To decide what number is the number of unpunctuated words in (T - text)
    (documented at ph_numupwords):
    (- TEXT_TY_BlobAccess({-by-reference:T}, UWORD_BLOB) -).
To decide what number is the number of lines in (T - text)
    (documented at ph_numlines):
    (- TEXT_TY_BlobAccess({-by-reference:T}, LINE_BLOB) -).
To decide what number is the number of paragraphs in (T - text)
    (documented at ph_numparas):
    (- TEXT_TY_BlobAccess({-by-reference:T}, PARA_BLOB) -).

To decide what text is character number (N - a number) in (T - text)
    (documented at ph_charnum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, CHR_BLOB) -).
To decide what text is word number (N - a number) in (T - text)
    (documented at ph_wordnum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, WORD_BLOB) -).
To decide what text is punctuated word number (N - a number) in (T - text)
    (documented at ph_pwordnum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, PWORD_BLOB) -).
To decide what text is unpunctuated word number (N - a number) in (T - text)
    (documented at ph_upwordnum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, UWORD_BLOB) -).
To decide what text is line number (N - a number) in (T - text)
    (documented at ph_linenum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, LINE_BLOB) -).
To decide what text is paragraph number (N - a number) in (T - text)
    (documented at ph_paranum):
    (- TEXT_TY_GetBlob({-new:text}, {-by-reference:T}, {N}, PARA_BLOB) -).

§36. The "substituted form of" is a technicality most Inform users never need to think about; as a one-off phrase, it may as well go here.

To decide what text is the substituted form of (T - text)
    (documented at ph_subform):
    (- TEXT_TY_SubstitutedForm({-new:text}, {-by-reference:T}) -).

§37. A common matching engine is used for matching plain text...

See test case BIP-TextReplacement.

Section 2 - Matching and Replacing

To decide if (T - text) exactly matches the text (find - text),
    case insensitively
    (documented at ph_exactlymatches):
    (- TEXT_TY_Replace_RE(CHR_BLOB,{-by-reference:T},{-by-reference:find},0,{phrase options},1) -).
To decide if (T - text) matches the text (find - text),
    case insensitively
    (documented at ph_matches):
    (- TEXT_TY_Replace_RE(CHR_BLOB,{-by-reference:T},{-by-reference:find},0,{phrase options}) -).
To decide what number is number of times (T - text) matches the text
    (find - text), case insensitively
    (documented at ph_nummatches):
    (- TEXT_TY_Replace_RE(CHR_BLOB,{-by-reference:T},{-by-reference:find},1,{phrase options}) -).

To replace the text (find - text) in (T - text) with (replace - text),
    case insensitively
    (documented at ph_replace):
    (- TEXT_TY_Replace_RE(CHR_BLOB, {-lvalue-by-reference:T}, {-by-reference:find},
        {-by-reference:replace}, {phrase options}); -).
To replace the word (find - text) in (T - text) with
    (replace - text)
    (documented at ph_replacewordin):
    (- TEXT_TY_ReplaceText(WORD_BLOB, {-lvalue-by-reference:T}, {-by-reference:find}, {-by-reference:replace}); -).
To replace the punctuated word (find - text) in (T - text)
    with (replace - text)
    (documented at ph_replacepwordin):
    (- TEXT_TY_ReplaceText(PWORD_BLOB, {-lvalue-by-reference:T}, {-by-reference:find}, {-by-reference:replace}); -).

To replace character number (N - a number) in (T - text)
    with (replace - text)
    (documented at ph_replacechar):
    (- TEXT_TY_ReplaceBlob(CHR_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To replace word number (N - a number) in (T - text)
    with (replace - text)
    (documented at ph_replaceword):
    (- TEXT_TY_ReplaceBlob(WORD_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To replace punctuated word number (N - a number) in (T - text)
    with (replace - text)
    (documented at ph_replacepword):
    (- TEXT_TY_ReplaceBlob(PWORD_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To replace unpunctuated word number (N - a number) in (T - text)
    with (replace - text)
    (documented at ph_replaceupword):
    (- TEXT_TY_ReplaceBlob(UWORD_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To replace line number (N - a number) in (T - text) with (replace - text)
    (documented at ph_replaceline):
    (- TEXT_TY_ReplaceBlob(LINE_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To replace paragraph number (N - a number) in (T - text) with (replace - text)
    (documented at ph_replacepara):
    (- TEXT_TY_ReplaceBlob(PARA_BLOB, {-lvalue-by-reference:T}, {N}, {-by-reference:replace}); -).
To decide what number is the first index of text match
    (documented at ph_textfirstindex):
    (- (match0_idx2 ~= 0) * (match0_idx + 1) -).
To decide what number is the last index of text match
    (documented at ph_textlastindex):
    (- match0_idx2 -).
To decide what number is the length of text match
    (documented at ph_textlength):
    (- (match0_idx2 - match0_idx) -).