MacLisp improved on the Lisp 1.5 notion of special variables and error handling. MacLisp also introduced the concept of functions that could take a variable number of arguments, macros, arrays, non-local dynamic exits, fast arithmetic, the first good Lisp compiler, and an emphasis on execution speed. This next sentence transplanted from a later paragraph about PSL. JonL says: My recollection was that it was closer to 100 than to 50. More likely, it had been received by about 100 sites, and was actually in continued usage by 50 at the end of the decade. KMP: Changed "was available" to "was in use" to compensate. (6-Dec-91)By the end of the 1970's, MacLisp was in use at over 50 sites. From about 1969 onward Jonl White was the dominant force behind Maclisp.For further information about Maclisp, see Maclisp Reference Manual, Revision 0 or The Revised Maclisp Manual.
Interlisp introduced many ideas into Lisp programming environments and methodology. One of the Interlisp ideas that influenced Common Lisp was an iteration construct implemented by Warren Teitelman that inspired the loop macro used both on the Lisp Machines and in MacLisp, and now in Common Lisp. For further information about Interlisp, see Interlisp Reference Manual.
Although the first implementations of Lisp were on the IBM 704 and the IBM 7090, later work focussed on the DEC PDP-6 and, later, PDP-10 computers, the latter being the mainstay of Lisp and artificial intelligence work at such places as Massachusetts Institute of Technology (MIT), Stanford University, and Carnegie Mellon University (CMU) from the mid-1960's through much of the 1970's. The PDP-10 computer and its predecessor the PDP-6 computer were, by design, especially well-suited to Lisp because they had 36-bit words and 18-bit addresses. This architecture allowed a cons cell to be stored in one word; single instructions could extract the car and cdr parts. The PDP-6 and PDP-10 had fast, powerful stack instructions that enabled fast function calling. But the limitations of the PDP-10 were evident by 1973: it supported a small number of researchers using Lisp, and the small, 18-bit address space ( 262,144 words) limited the size of a single program. One response to the address space problem was the Lisp Machine, a special-purpose computer designed to run Lisp programs. The other response was to use general-purpose computers with address spaces larger than 18 bits, such as the DEC VAX and the S-1 Mark IIA. For further information about S-1 Common Lisp, see “S-1 Common Lisp Implementation.”
The Lisp machine concept was developed in the late 1960's. In the early 1970's, Peter Deutsch, working with Daniel Bobrow, implemented a Lisp on the Alto, a single-user minicomputer, using microcode to interpret a byte-code implementation language. Shortly thereafter, Richard Greenblatt began work on a different hardware and instruction set design at MIT. Although the Alto was not a total success as a Lisp machine, a dialect of Interlisp known as Interlisp-D became available on the D-series machines manufactured by Xerox—the Dorado, Dandelion, Dandetiger, and Dove (or Daybreak). An upward-compatible extension of MacLisp called Lisp Machine Lisp became available on the early MIT Lisp Machines. Commercial Lisp machines from Xerox, Lisp Machines (LMI), and Symbolics were on the market by 1981. For further information about Lisp Machine Lisp, see Lisp Machine Manual.
During the late 1970's, Lisp Machine Lisp began to expand towards a much fuller language. Sophisticated lambda lists, setf, multiple values, and structures like those in Common Lisp are the results of early experimentation with programming styles by the Lisp Machine group. Jonl White and others migrated these features to MacLisp. Around 1980, Scott Fahlman and others at CMU began work on a Lisp to run on the Scientific Personal Integrated Computing Environment (SPICE) workstation. One of the goals of the project was to design a simpler dialect than Lisp Machine Lisp.
The Macsyma group at MIT began a project during the late 1970's called the New Implementation of Lisp (NIL) for the VAX, which was headed by White. One of the stated goals of the NIL project was to fix many of the historic, but annoying, problems with Lisp while retaining significant compatibility with MacLisp. At about the same time, a research group at Stanford University and Lawrence Livermore National Laboratory headed by Richard P. Gabriel began the design of a Lisp to run on the S-1 Mark IIA supercomputer. S-1 Lisp, never completely functional, was the test bed for adapting advanced compiler techniques to Lisp implementation. Eventually the S-1 and NIL groups collaborated. For further information about the NIL project, see “NIL—A Perspective.”
The first efforts towards Lisp standardization were Standard Lisp and Portable Standard Lisp (PSL). In 1969, Anthony Hearn and Martin Griss defined Standard Lisp---a subset of Lisp~1.5 and other dialects---to transport REDUCE, a symbolic algebra system. PSL was designed to provide more control over the environment and the compiler. At the end of the 1970's, PSL ran on about a dozen different computers. PSL and Franz Lisp---a MacLisp-like dialect for Unix machines---were the first examples of widely available Lisp dialects on multiple hardware platforms. MacLisp was available at over 50 sites. For further information about Standard Lisp, see ``{\StandardLispReport}.'' Sandra: This paragraph contains inaccuracies and is not well-organized. Suggested rewrite [taken -kmp] below.
The first effort towards Lisp standardization was made in 1969, when Anthony Hearn and Martin Griss at the University of Utah defined Standard Lisp—a subset of Lisp 1.5 and other dialects—to transport REDUCE, a symbolic algebra system. During the 1970's, the Utah group implemented first a retargetable optimizing compiler for Standard Lisp, and then an extended implementation known as Portable Standard Lisp (PSL). By the mid 1980's, PSL ran on about a dozen kinds of computers. For further information about Standard Lisp, see “Standard LISP Report.”
PSL and Franz Lisp—a MacLisp-like dialect for Unix machines—were the first examples of widely available Lisp dialects on multiple hardware platforms.
One of the most important developments in Lisp occurred during the second half of the 1970's: Scheme. Scheme, designed by Gerald J. Sussman and Guy L. Steele Jr., is a simple dialect of Lisp whose design brought to Lisp some of the ideas from programming language semantics developed in the 1960's. Sussman was one of the prime innovators behind many other advances in Lisp technology from the late 1960's through the 1970's. The major contributions of Scheme were lexical scoping, lexical closures, first-class continuations, and simplified syntax (no separation of value cells and function cells). Some of these contributions made a large impact on the design of Common Lisp. For further information about Scheme, see IEEE Standard for the Scheme Programming Language or “Revised Report on the Algorithmic Language Scheme.”
In the late 1970's object-oriented programming concepts started to make a strong impact on Lisp. At MIT, certain ideas from Smalltalk made their way into several widely used programming systems. Flavors, an object-oriented programming system with multiple inheritance, was developed at MIT for the Lisp machine community by Howard Cannon and others. At Xerox, the experience with Smalltalk and Knowledge Representation Language (KRL) led to the development of Lisp Object Oriented Programming System (LOOPS) and later Common LOOPS. For further information on Smalltalk, see Smalltalk-80: The Language and its Implementation. For further information on Flavors, see Flavors: A Non-Hierarchical Approach to Object-Oriented Programming.
These systems influenced the design of the Common Lisp Object System (CLOS). CLOS was developed specifically for this standardization effort, and was separately written up in “Common Lisp Object System Specification.” However, minor details of its design have changed slightly since that publication, and that paper should not be taken as an authoritative reference to the semantics of the object system as described in this document.
In 1980 Symbolics and LMI were developing Lisp Machine Lisp; stock-hardware implementation groups were developing NIL, Franz Lisp, and PSL; Xerox was developing Interlisp; and the SPICE project at CMU was developing a MacLisp-like dialect of Lisp called SpiceLisp.
In April 1981, after a DARPA-sponsored meeting concerning the splintered Lisp community, Symbolics, the SPICE project, the NIL project, and the S-1 Lisp project joined together to define Common Lisp. Initially spearheaded by White and Gabriel, the driving force behind this grassroots effort was provided by Fahlman, Daniel Weinreb, David Moon, Steele, and Gabriel. Common Lisp was designed as a description of a family of languages. The primary influences on Common Lisp were Lisp Machine Lisp, MacLisp, NIL, S-1 Lisp, Spice Lisp, and Scheme. Common Lisp: The Language is a description of that design. Its semantics were intentionally underspecified in places where it was felt that a tight specification would overly constrain Common Lisp research and use. Per X3J13. -kmp 05-Oct-93 Between 1984 and 1989, \clisp\ became a de facto standard.
In 1986 X3J13 was formed as a technical working group to produce a draft for an ANSI Common Lisp standard. Because of the acceptance of Common Lisp, the goals of this group differed from those of the original designers. These new goals included stricter standardization for portability, an object-oriented programming system, a condition system, iteration facilities, and a way to handle large character sets. To accommodate those goals, a new language specification, this document, was developed.
This is a reference document, not a tutorial document. Where possible and convenient, the order of presentation has been chosen so that the more primitive topics precede those that build upon them; however, linear readability has not been a priority.
This document is divided into chapters by topic. Any given chapter might contain conceptual material, dictionary entries, or both.
Defined names within the dictionary portion of a chapter are grouped in a way that brings related topics into physical proximity. Many such groupings were possible, and no deep significance should be inferred from the particular grouping that was chosen. To see defined names grouped alphabetically, consult the index. For a complete list of defined names, see Section 1.9 (Symbols in the COMMON-LISP Package).
In order to compensate for the sometimes-unordered portions of this document, a glossary has been provided; see Chapter 26 (Glossary). The glossary provides connectivity by providing easy access to definitions of terms, and in some cases by providing examples or cross references to additional conceptual material.
For information about notational conventions used in this document, see Section 1.4 (Definitions).
For information about conformance, see Section 1.5 (Conformance).
For information about extensions and subsets, see Section 1.6 (Language Extensions) and Section 1.7 (Language Subsets).
For information about how programs in the language are parsed by the Lisp reader, see Chapter 2 (Syntax).
For information about how programs in the language are compiled and executed, see Chapter 3 (Evaluation and Compilation).
For information about data types, see Chapter 4 (Types and Classes). Not all types and classes are defined in this chapter; many are defined in chapter corresponding to their topic–for example, the numeric types are defined in Chapter 12 (Numbers). For a complete list of standardized types, see Figure 4–2.
For information about general purpose control and data flow, see Chapter 5 (Data and Control Flow) or Chapter 6 (Iteration).
JonL wanted this added. See entry for HASH-TABLE. -kmp 5-Feb-92
DefinitionsThis section contains notational conventions and definitions of terms used in this manual.
The following notational conventions are used throughout this document.
========================================
Fonts are used in this document to convey information.
Denotes a formal term whose meaning is defined in the Glossary. When this font is used, the Glossary definition takes precedence over normal English usage.
Sometimes a glossary term appears subscripted, as in “whitespace2.” Such a notation selects one particular Glossary definition out of several, in this case the second. The subscript notation for Glossary terms is generally used where the context might be insufficient to disambiguate among the available definitions.
Denotes the introduction of a formal term locally to the current text. There is still a corresponding glossary entry, and is formally equivalent to a use of “name,” but the hope is that making such uses conspicuous will save the reader a trip to the glossary in some cases.
name
Denotes a symbol in the common-lisp package. For information about case conventions, see Section 1.4.1.4.1 (Case in Symbols).
name
Denotes a sample name or piece of code that a programmer might write in Common Lisp.
This font is also used for certain standardized names that are not names of external symbols of the common-lisp package, such as keywords1, package names, and loop keywords.
Denotes the name of a parameter or value.
In some situations the notation “⟪name⟫” (i.e., the same font, but with surrounding “angle brackets”) is used instead in order to provide better visual separation from surrounding characters. These “angle brackets” are metasyntactic, and never actually appear in program input or output.
========================================
This specification uses an extended Backus Normal Form (BNF) to describe the syntax of Common Lisp macro forms and special forms. This section discusses the syntax of BNF expressions.
An expression of this form appears whenever a list of elements is to be spliced into a larger structure and the elements can appear in any order. The symbol represents a description of the syntax of some number of syntactic elements to be spliced; that description must be of the form
where each can be of the form or of the form * or of the form 1. The expression ⟦⟧ means that a list of the form
is spliced into the enclosing expression, such that if and , then either or , where for some , is of the form *. Added to accomodate new LOOP BNF. -kmp 1-May-93Furthermore, for each that is of the form 1, that element is required to appear somewhere in the list to be spliced.
For example, the expression
(x ⟦A | B* | C⟧ y)
means that at most one A, any number of B's, and at most one C can occur in any order. It is a description of any of these:
(x y) (x B A C y) (x A B B B B B C y) (x C B A B B B y)
but not any of these:
(x B B A A C C y) (x C B C y)
In the first case, both A and C appear too often, and in the second case C appears too often.
I added this notation to make the LOOP bvl easier to specify. -kmp 29-Apr-93
The notation ⟦ | | ⟧+ adds the additional restriction that at least one item from among the possible choices must be used. For example:
(x ⟦A | B* | C⟧+ y)
means that at most one A, any number of B's, and at most one C can occur in any order, but that in any case at least one of these options must be selected. It is a description of any of these:
(x B y) (x B A C y) (x A B B B B B C y) (x C B A B B B y)
but not any of these:
(x y) (x B B A A C C y) (x C B C y)
In the first case, no item was used; in the second case, both A and C appear too often; and in the third case C appears too often.
Also, the expression:
(x ⟦A1 | B1 | C⟧ y)
can generate exactly these and no others:
(x A B C y) (x A C B y) (x A B y) (x B A C y) (x B C A y) (x B A y) (x C A B y) (x C B A y)
If O is a non-terminal symbol, the right-hand side of its definition is substituted for the entire expression ↓O. For example, the following BNF is equivalent to the BNF in the previous example:
(x ⟦↓O⟧ y)
A | B* | C
case keyform {↓normal-clause}* [↓otherwise-clause] → {result}*
ccase keyplace {↓normal-clause}* → {result}*
ecase keyform {↓normal-clause}* → {result}*
Here the term “clause” might appear to be “dead” in that it is not used in the BNF. However, the purpose of the BNF is not just to guide parsing, but also to define useful terms for reference in the descriptive text which follows. As such, the term “clause” might appear in text that follows, as shorthand for “normal-clause or otherwise-clause.”
========================================
This indicates evaluation. For example:
(+ 4 5) → 9This means that the result of evaluating the form
(+ 4 5) is 9.
!!! Are the first two of these notations still used? Maybe remove...If a form returns multiple values, those values might be shown separated by spaces, line breaks, or commas. For example:
(truncate 7 5) → 1 2 (truncate 7 5) → 1 2 (truncate 7 5) → 1, 2
Each of the above three examples is equivalent, and specifies that (truncate 7 5) returns two values, which are 1 and 2.
Some conforming implementations actually type an arrow (or some other indicator) before showing return values, while others do not.
The notation “OR→” is used to denote one of several possible alternate results. The example
(char-name #\a) → NIL OR→ "LOWERCASE-a" OR→ "Small-A" OR→ "LA01"
indicates that nil, "LOWERCASE-a", "Small-A", "LA01" are among the possible results of (char-name #\a)—each with equal preference. Unless explicitly specified otherwise, it should not be assumed that the set of possible results shown is exhaustive. Formally, the above example is equivalent to
(char-name #\a) → implementation-dependent
but it is intended to provide additional information to illustrate some of the ways in which it is permitted for implementations to diverge.
The notation “NOT→” is used to denote a result which is not possible. This might be used, for example, in order to emphasize a situation where some anticipated misconception might lead the reader to falsely believe that the result might be possible. For example,
(function-lambda-expression
(funcall #'(lambda (x) #'(lambda () x)) nil))
→ NIL, true, NIL
OR→ (LAMBDA () X), true, NIL
NOT→ NIL, false, NIL
NOT→ (LAMBDA () X), false, NIL
1.2.3 3
This indicates code equivalence. For example:
(gcd x (gcd y z)) ≡ (gcd (gcd x y) z)This means that the results and observable side-effects of evaluating the form
(gcd x (gcd y z)) are always the same as the results and observable side-effects of (gcd (gcd x y) z) for any x, y, and z.
!!! Need to expand on the definition of observable side-effects.
Common Lisp specifies input and output with respect to a non-interactive stream model. The specific details of how interactive input and output are mapped onto that non-interactive model are implementation-defined.
For example, conforming implementations are permitted to differ in issues of how interactive input is terminated. For example, the function read terminates when the final delimiter is typed on a non-interactive stream. In some implementations, an interactive call to read returns as soon as the final delimiter is typed, even if that delimiter is not a newline. In other implementations, a final newline is always required. In still other implementations, there might be a command which “activates” a buffer full of input without the command itself being visible on the program's input stream.
In the examples in this document, the notation “⊳” precedes lines where interactive input and output occurs. Within such a scenario, “this notation” notates user input.
For example, the notation
(+ 1 (print (+ (sqrt (read)) (sqrt (read))))) ⊳ 9 16 ⊳ 7 → 8
shows an interaction in which “(+ 1 (print (+ (sqrt (read)) (sqrt (read)))))” is a form to be evaluated, “9 16 ” is interactive input, “7” is interactive output, and “8” is the value yielded from the evaluation.
The use of this notation is intended to disguise small differences in interactive input and output behavior between implementations.
Sometimes, the non-interactive stream model calls for a newline. How that newline character is interactively entered is an implementation-defined detail of the user interface, but in that case, either the notation “⟨Newline⟩” or “↩” might be used.
(progn (format t "~&Who? ") (read-line)) ⊳ Who? Fred, Mary, and Sally↩ → "Fred, Mary, and Sally", false
========================================
----------------------------------------
While case is significant in the process of interning a symbol, the Lisp reader, by default, attempts to canonicalize the case of a symbol prior to interning; see Section 23.1.2 (Effect of Readtable Case on the Lisp Reader). As such, case in symbols is not, by default, significant. Throughout this document, except as explicitly noted otherwise, the case in which a symbol appears is not significant; that is, HELLO, Hello, hElLo, and hello are all equivalent ways to denote a symbol whose name is "HELLO".
The characters backslash and vertical-bar are used to explicitly quote the case and other parsing-related was "attributes" but now that attributes has formaly meaning,
not sure if it's too limiting here, so use a general term. -kmp 26-Jan-92aspects of characters. As such, the notations |hello| and \h\e\l\l\o are equivalent ways to refer to a symbol whose name is "hello", and which is distinct from any symbol whose name is "HELLO".
The symbols that correspond to Common Lisp defined names have uppercase names even though their names generally appear in lowercase in this document.
----------------------------------------
----------------------------------------
(item1 item2 . tail)
means that tail represents a list of objects at the end of a list. For example,
(A B C . (D E F))
is notationally equivalent to:
(A B C D E F)
Although dot is a valid constituent character in a symbol, no standardized symbols contain the character dot, so a period that follows a reference to a symbol at the end of a sentence in this document should always be interpreted as a period and never as part of the symbol's name. For example, within this document, a sentence such as “This sample sentence refers to the symbol car.” confusion about fonts below made more consistent w/previous section
on symbol names --sjl 13 Mar 1992
refers to a function named ``\funref{car}'' with three letters in its name,
and never to a four-letter symbol ``\funref{car.}''refers to a symbol whose name is "CAR" (with three letters), and never to a four-letter symbol "CAR."
----------------------------------------
nil has a variety of meanings. It is a symbol in the common-lisp package with the name "NIL", it is boolean (and generalized boolean) false, it is the empty list, and it is the name of the empty type (a subtype of all types).
Within Common Lisp, nil can be notated interchangeably as either NIL or (). By convention, the choice of notation offers a hint as to which of its many roles it is playing.
| For Evaluation? | Notation | Typically Implied Role |
| Yes | nil | use as a boolean. |
| Yes | 'nil | use as a symbol. |
| Yes | '() | use as an empty list |
| No | nil | use as a symbol or boolean. |
| No | () | use as an empty list. |
Figure 1–1. Notations for NIL
Within this document only, nil is also sometimes notated as false to emphasize its role as a boolean.
For example:
(print ()) ;avoided (defun three nil 3) ;avoided '(nil nil) ;list of two symbols '(() ()) ;list of empty lists (defun three () 3) ;Emphasize empty parameter list. (append '() '()) → () ;Emphasize use of empty lists (not nil) → true ;Emphasize use as Boolean false (get 'nil 'color) ;Emphasize use as a symbol
A function is sometimes said to “be false” or “be true” in some circumstance. Since no function object can be the same as nil and all function objects represent true when viewed as booleans, it would be meaningless to say that the function was literally false and uninteresting to say that it was literally true. Instead, these phrases are just traditional alternative ways of saying that the function “returns false” or “returns true,” respectively.
========================================
A designator is an object that denotes another object.
!!! RPG: Not clear. Rewrite.Where a parameter of an operator is described as a designator, the description of the operator is written in a way that assumes that the value of the parameter is the denoted object; that is, that the parameter is already of the denoted type. (The specific nature of the object denoted by a “⟪type⟫ designator” or a “designator for a ⟪type⟫” can be found in the Glossary entry for “⟪type⟫ designator.”)
For example, “nil” and “the value of *standard-output*” are operationally indistinguishable as stream designators. Similarly, the symbol foo and the string "FOO" are operationally indistinguishable as string designators.
Except as otherwise noted, in a situation where the denoted object might be used multiple times, it is implementation-dependent whether the object is coerced only once or whether the coercion occurs each time the object must be used.
For example, mapcar receives a function designator as an argument, and its description is written as if this were simply a function. In fact, it is implementation-dependent whether the function designator is coerced right away or whether it is carried around internally in the form that it was given as an argument and re-coerced each time it is needed. In most cases, conforming programs cannot detect the distinction, but there are some pathological situations (particularly those involving self-redefining or mutually-redefining functions) which do conform and which can detect this difference. The following program is a conforming program, but might or might not have portably correct results, depending on whether its correctness depends on one or the other of the results:
(defun add-some (x) (defun add-some (x) (+ x 2)) (+ x 1)) → ADD-SOME (mapcar 'add-some '(1 2 3 4)) → (2 3 4 5) OR→ (2 4 5 6)
In a few rare situations, there may be a need in a dictionary entry to refer to the object that was the original designator for a parameter. Since naming the parameter would refer to the denoted object, the phrase “the ⟪parameter-name⟫ designator” can be used to refer to the designator which was the argument from which the value of ⟪parameter-name⟫ was computed.
========================================
When a word having no pre-attached semantics is required (e.g., in an example), it is common in the Lisp community to use one of the words “foo,” “bar,” “baz,” and “quux.” For example, in
(defun foo (x) (+ x 1))the use of the name
foo is just a shorthand way of saying “please substitute your favorite name here.”
These nonsense words have gained such prevalance of usage, that it is commonplace for newcomers to the community to begin to wonder if there is an attached semantics which they are overlooking—there is not.
!!! Barmar thinks \secref\InterpretingDictionaryEntries should move to someplace around here.
Situations in which errors might, should, or must be signaled are described in the standard. The wording used to describe such situations is intended to have precise meaning. The following list is a glossary of those meanings.
This is code processed with the safety optimization at its highest setting (3). safety is a lexical property of code. The phrase “the function F should signal an error” means that if F is invoked from code processed with the highest safety optimization, an error is signaled. It is implementation-dependent whether F or the calling code signals the error.
This is code processed with lower safety levels.
Unsafe code might do error checking. Implementations are permitted to treat all code as safe code all the time.
1.2.4 6 1.2.4 9 1.2.4 10
This means that an error is signaled in both safe and unsafe code. Conforming code may rely on the fact that the error is signaled in both safe and unsafe code. Every implementation is required to detect the error in both safe and unsafe code. For example, “an error is signaled if unexport is given a symbol not accessible in the current package.”
If an explicit error type is not specified, the default is error.
This means that an error is signaled in safe code, and an error might be signaled in unsafe code. Conforming code may rely on the fact that the error is signaled in safe code. Every implementation is required to detect the error at least in safe code. When the error is not signaled, the “consequences are undefined” (see below). For example, “+ should signal an error of type type-error if any argument is not of type number.”
This is similar to “should be signaled” except that it does not imply that `extra effort' has to be taken on the part of an operator to discover an erroneous situation if the normal action of that operator can be performed successfully with only `lazy' checking. An implementation is always permitted to signal an error, but even in safe code, it is only required to signal the error when failing to signal it might lead to incorrect results. In unsafe code, the consequences are undefined.
For example, defining that “find should be prepared to signal an error of type type-error if its second argument is not a proper list” does not imply that an error is always signaled. The form
(find 'a '(a b . c))
must either signal an error of type type-error in safe code, else return A. In unsafe code, the consequences are undefined. By contrast,
(find 'd '(a b . c))
must signal an error of type type-error in safe code. In unsafe code, the consequences are undefined. Also,
(find 'd '#1=(a b . #1#))
in safe code might return nil (as an implementation-defined extension), might never return, or might signal an error of type type-error. In unsafe code, the consequences are undefined.
Typically, the “should be prepared to signal” terminology is used in type checking situations where there are efficiency considerations that make it impractical to detect errors that are not relevant to the correct operation of the operator.
This means that the consequences are unpredictable but harmless. Implementations are permitted to specify the consequences of this situation. No conforming code may depend on the results or effects of this situation, and all conforming code is required to treat the results and effects of this situation as unpredictable but harmless. For example, “if the second argument to shared-initialize specifies a name that does not correspond to any slots accessible in the object, the results are unspecified.”
1.2.4 4
This means that the consequences are unpredictable. The consequences may range from harmless to fatal. No conforming code may depend on the results or effects. Conforming code must treat the consequences as unpredictable. In places where the words “must,” “must not,” or “may not” are used, then “the consequences are undefined” if the stated requirement is not met and no specific consequence is explicitly stated. An implementation is permitted to signal an error in this case.
For example: “Once a name has been declared by defconstant to be constant, any further assignment or binding of that variable has undefined consequences.”
This means that the situation has undefined consequences; however, if an error is signaled, it is of the specified type. For example, “open might signal an error of type file-error.”
This means that only the number and nature of the return values of a form are not specified. However, the issue of whether or not any side-effects or transfer of control occurs is still well-specified.
A program can be well-specified even if it uses a function whose returns values are unspecified. For example, even if the return values of some function F are unspecified, an expression such as (length (list (F))) is still well-specified because it does not rely on any particular aspect of the value or values returned by F.
This means that the situation has undefined consequences; however, a conforming implementation is free to treat the situation in a more specific way. For example, an implementation might define that an error is signaled, or that an error should be signaled, or even that a certain well-defined non-error behavior occurs.
1.2.4 5No conforming code may depend on the consequences of such a situation; all conforming code must treat the consequences of the situation as undefined. Implementations are required to document how the situation is treated.
For example, “implementations may be extended to define other type specifiers to have a corresponding class.”
This means that in this situation implementations are permitted to define unambiguous extensions to the syntax of the form being described. No conforming code may depend on this extension. Implementations are required to document each such extension. All conforming code is required to treat the syntax as meaningless. The standard might disallow certain extensions while allowing others. For example, “no implementation is free to extend the syntax of defclass.”
This means that implementations are encouraged to issue a warning if the context is appropriate (e.g., when compiling). However, a conforming implementation is not required to issue a warning.
This means that a warning might be issued, as described in \funref{warn}, in both safe and unsafe code when the situation is detected. \term{Conforming code} can rely on the fact that a warning will be issued in both safe and unsafe code when the situation is detected. Every implementation is required to detect this situation in both safe and unsafe code when the situation is detected. The presence of a warning will in no way alter the value returned by the \term{form} that caused the situation to occur. For example, ``a warning is issued if a declaration specifier is not one of those defined in the description of \misc{declare} and has not been declared in a \declref{declaration} declaration.'' \itemitem{\b{A warning should be issued}} This means that a warning might be issued. \term{Conforming code} must not rely on the fact that a warning will be issued. If the situation is detected, a warning might be issued. The presence of a warning will in no way alter the value returned by the \term{form} that caused the situation to occur. For example, ``a warning should be issued if a variable declared \declref{ignore} is ever referenced or is also declared \declref{special}, or if a variable is lexical, never referenced, and not declared \declref{ignore}.''
A lot of this seems to be just rationale. Does it really need to be included here? --sjl 13 Mar 92
Front matter and back matter, such as the “Table of Contents,” “Index,” “Figures,” “Credits,” and “Appendix” are not considered formally part of this standard, so that we retain the flexibility needed to update these sections even at the last minute without fear of needing a formal vote to change those parts of the document. These items are quite short and very useful, however, and it is not recommended that they be removed even in an abridged version of this document.
Within the concept sections, subsections whose names begin with the words “Note” or “Notes” or “Example” or “Examples” are provided for illustration purposes only, and are not considered part of the standard.
An attempt has been made to place these sections last in their parent section, so that they could be removed without disturbing the contiguous numbering of the surrounding sections in order to produce a document of smaller size.
Likewise, the “Examples” and “Notes” sections in a dictionary entry are not considered part of the standard and could be removed if necessary.
Nevertheless, the examples provide important clarifications and consistency checks for the rest of the material, and such abridging is not recommended unless absolutely unavoidable.
The dictionary entry for each defined name is partitioned into sections. Except as explicitly indicated otherwise below, each section is introduced by a label identifying that section. The omission of a section implies that the section is either not applicable, or would provide no interesting information.
This section defines the significance of each potential section in a dictionary entry.
For a variable, anything that can affect the value of the variable including functions that bind or assign it.
I added the first part of this sentence as editorial discretion since I believe we strongly feel that this is not specified otherwise, but we want to avoid an unexpected conflict in case it is. -kmp 9-May-94Except as explicitly specified otherwise, Added per X3J13 at May 4-5, 1994 meeting. -kmp 9-May-94the consequences are undefined if these type restrictions are violated.
It is permissible for other (implementation-defined) classes to appear in the implementation's class precedence list for the class.
It is permissible for either standard-object or structure-object to appear in the implementation's class precedence list; for details, see Section 4.2.2 (Type Relationships).
Except as explicitly indicated otherwise somewhere in this specification, no additional standardized classes may appear in the implementation's class precedence list.
By definition of the relationship between classes and types, the classes listed in this section are also supertypes of the type denoted by the class.
The atomic type specifiers are those defined names listed in Figure 4–2. Such dictionary entries are of kind “Class,” “Condition Type,” “System Class,” or “Type.” A description of how to interpret a symbol naming one of these types or classes as an atomic type specifier is found in the “Description” section of such dictionary entries.
The compound type specifiers are those defined names listed in Figure 4–3. Such dictionary entries are of kind “Class,” “System Class,” “Type,” or “Type Specifier.” A description of how to interpret as a compound type specifier a list whose car is such a symbol is found in the “Compound Type Specifier Kind,” “Compound Type Specifier Syntax,” “Compound Type Specifier Arguments,” and “Compound Type Specifier Description” sections of such dictionary entries.
A “specializing” type specifier is one that describes a subtype by restricting the type of one or more components of the type, such as element type or complex part type.
A “predicating” type specifier is one that describes a subtype containing only those objects that satisfy a given predicate.
A “combining” type specifier is one that describes a subtype in a compositional way, using combining operations (such as “and,” “or,” and “not”) on other types.
Whether or not the type is acceptable as an atomic type specifier is not represented here; see Section 1.4.4.6 (Dictionary Entries for Type Specifiers).
F (x class) (y t) &optional z &key k
This signature indicates that this method on the generic function F has two required parameters: x, which must be a generalized instance of the class class; and y, which can be any object (i.e., a generalized instance of the class t). In addition, there is an optional parameter z and a keyword parameter k. This signature also indicates that this method on F is a primary method and has no qualifiers.
For each parameter, the argument supplied must be in the intersection of the type specified in the description of the corresponding generic function and the type given in the signature of some method (including not only those methods defined in this specification, but also implementation-defined or user-defined methods in situations where the definition of such methods is permitted).
In large print at left, the defined name appears; if more than one defined name is to be described by the entry, all such names are shown separated by commas.
In somewhat smaller italic print at right is an indication of what kind of dictionary entry this is. Possible values are:
This list believed correct as of 23-Oct-91 -kmp
This is a class.
This is a subtype of type condition. We don't need to constrain how implementations define condition types.
--sjl 13 Mar 92
\term{Condition types} are defined with
\macref{define-condition}, not \macref{defclass}.
This is a constant variable.
This is a declaration identifier.
This is a function.
This is a function that is defined only lexically within the scope of some other macro form.
This is a macro that is defined only lexically within the scope of some other macro form.
This is a macro.
This is a restart.
This is a special operator.
This is a standard generic function.
This is a symbol that is specially recognized in some particular situation, such as the syntax of a macro.
This is like class, but it identifies a class that is potentially a built-in class. (No class is actually required to be a built-in class.)
This is an atomic type specifier, and depending on information for each particular entry, may subject to form other type specifiers.
This is a defined name that is not an atomic type specifier, but that can be used in constructing valid type specifiers.
This is a dynamic variable.
!!! Is this needed? Mail sent to `barmar' and `barrett' with subject "supertypes" asking for opinions. -kmp 10-Feb-92In implementations where there is a corresponding class, the order of the classes in the class precedence list is consistent with the order presented in this section.
For example, an operator description might say:
F x y &optional z &key k
This description indicates that the function F has two required parameters, x and y. In addition, there is an optional parameter z and a keyword parameter k.
For macros and special operators, syntax is given in modified BNF notation; see Section 1.4.1.2 (Modified BNF Syntax). For functions a lambda list is given. Added per Barmar:In both cases, however, the outermost parentheses are omitted, and default value information is omitted.
file-position stream → position
file-position stream position-spec → success-p
is operationally equivalent to this line:
file-position stream &optional position-spec → result
and differs only in that it provides on opportunity to introduce different names for parameter and values for each case. The separated (multi-line) notation is used when an operator is overloaded in such a way that the parameters are used in different ways depending on how many arguments are supplied (e.g., for the function /) or the return values are different in the two cases (e.g., for the function file-position).
For example, given a syntax description such as:
F &rest arguments
it is appropriate to refer either to the rest parameter named arguments by name, or to one of its elements by speaking of “an argument,” “some argument,” “each argument” etc..
F &rest arguments+
means the same as
F &rest arguments
but introduces the additional requirement that there be at least one argument.
F a b c → x
indicates that F is an operator that has three required parameters (i.e., a, b, and c) and that returns one value (i.e., x). If more than one value is returned by an operator, the names of the values are separated by commas, as in:
F a b c → x, y, z
indicates that F is an operator that accepts no arguments and returns no values.
F a b c →
A given “Declaration” might appear in a declaration (i.e., a declare expression), a proclamation (i.e., a declaim or proclaim form), or both.
I added the first part of this sentence as editorial discretion since I believe we strongly feel that this is not specified otherwise, but we want to avoid an unexpected conflict in case it is. -kmp 9-May-94Except as explicitly specified otherwise, Added per X3J13 at May 4-5, 1994 meeting. -kmp 9-May-94the consequences are undefined if this type restriction is violated.
ComplianceThis standard presents the syntax and semantics to be implemented by a conforming implementation (and its accompanying documentation). In addition, it imposes requirements on conforming programs.
A conforming implementation shall accept all features (including deprecated features) of the language specified in this standard, with the meanings defined in this standard.
A conforming implementation shall not require the inclusion of substitute or additional language elements in code in order to accomplish a feature of the language that is specified in this standard.
In addition, a conforming implementation is encouraged (but not required) to document items in this standard that are identified as implementation-dependent, although in some cases such documentation might simply identify the item as “undefined.”
foo is a member of a set of functions that operate on numbers. Suppose that one passage states that an error must be signaled if any function in is ever given an argument of 17. Suppose that an apparently conflicting passage states that the consequences are undefined if foo receives an argument of 17. Then the second passage (the one specifically about foo) would dominate because the description of the situational context is the most specific, and it would not be required that foo signal an error on an argument of 17 even though other functions in the set would be required to do so.
in Exceptional Situations}
If the implementation conforms with some but not all of the requirements of this standard, then the conformance statement shall be
Code conforming with the requirements of this standard shall adhere to the following:
Rewritten per X3J13 and Boyer/Kaufmann/Moore #2 (second public review). -kmp 9-May-94 \itemitem{2.} \term{Conforming code} shall not rely on any particular interpretation of \term{implementation-dependent} features.
This material is all covered in more detail in chapter 3 (in the section on constraints on externalizable objects). There is no need to repeat the material here. --sjl 13 Mar 92 \beginsubsubsection{Conforming Programs in Files} A \term{conforming program} whose source text is in a \term{compilation unit} must satisfy the following constraints: \beginlist \item{\bull} Any \term{top level form} in a \term{file} that alters \thevalueof{*package*} at compile time must also alter it at load time to a \term{similar} \term{package}. \item{\bull} If the first \term{non-atomic} \term{top level form} in the \term{file} is not a call to \macref{in-package}, then the \term{current package} at load time must be a \term{similar} \term{package} to the one in effect at compile time. \item{\bull} Any \term{symbol} in the source text \term{accessible} in the \term{current package} at compile time and whose \term{home package} is the package $P$ must be \term{accessible} in the \term{current package} at load time and must also be \term{accessible} in a package \term{similar} to $P$. \item{\bull} Any \term{symbol} in the source text that is an \term{external symbol} of the package $P$ at compile time must be an \term{external symbol} of package \term{similar} to $P$ at load time. \endlist In \term{situations} in which any of these conditions does not hold, an \term{implementation} might signal an error or might extend \clisp\ to handle the situation. \endsubsubsection%{Conforming Programs in Files}
Moved to its own section (see below) per Dalton #1 (1st Public Review) by X3J13 vote May 4-5, 1994 (after 2nd public review). -kmp 9-May-94 !!! Barmar wonders if this is really the right place for the next sentence. \term{Portable code} is written using only \term{standard characters}.
Informally, the basic rules for conformance are as follows: 1. Conforming code is defined in terms of its structure, and not in terms of its results and side effects. 2. Conforming code is written using only the syntax specified in the standard, or syntax defined using the syntax extension mechanisms (macros and reader macros) specified in the standard. 3. Conforming code is written in only the sequence specified in the standard. 4. Conforming code is written using only the functions, macros, special forms, variables, and constants specified in the standard. 5. Conforming implementations provide the functionality and behavior specified in the standard. The definitions of all names and syntax that aren't specified in the standard and aren't provided by the implementation must accompany the code. 6. Conformance is not machine-checkable.
Conforming code may run in all conforming implementations, but might have allowable implementation-defined behavior that makes it non-portable code. Insofar as we allow options in the standard this will be true.For example, the following are examples of forms that are conforming, but that might return different values in different implementations:
(evenp most-positive-fixnum) → implementation-dependent (random) → implementation-dependent (> lambda-parameters-limit 93) → implementation-dependent (char-name #\A) → implementation-dependent
The following is added to clarify a question from RWK. Mail sent to him and Quinquevirate noting that this interpretation was made. -kmp 10-Apr-91
Use of #+ and #- does not automatically disqualify a program from being conforming. A program which uses #+ and #- is considered conforming if there is no set of features in which the program would not be conforming. Of course, conforming programs are not necessarily working programs. The following program is conforming:
(defun foo () #+ACME (acme:initialize-something) (print 'hello-there))
However, this program might or might not work, depending on whether the presence of the feature ACME really implies that a function named acme:initialize-something is present in the environment. In effect, using #+ or #- in a conforming program means that the variable *features* becomes just one more piece of input data to that program. Like any other data coming into a program, the programmer is responsible for assuring that the program does not make unwarranted assumptions on the basis of input data.
A language extension is any documented implementation-defined behavior of a defined name in this standard that varies from the behavior described in this standard, or a documented consequence of a situation that the standard specifies as undefined, unspecified, or extendable by the implementation. For example, if this standard says that “the results are unspecified,” an extension would be to specify the results.
Reviewer: Barmar: This contradicts previous definitions of conforming code. If the correct behavior of a program depends on the results provided by an extension, only implementations with the same extension will execute the program correctly. Note that such a program might be non-conforming. Also, if this standard says that “an implementation may be extended,” a conforming, but possibly non-portable, program can be written using an extension.
An implementation can have extensions, provided they do not alter the behavior of conforming code and provided they are not explicitly prohibited by this standard.
The term “extension” refers only to extensions available upon startup. An implementation is free to allow or prohibit redefinition of an extension.
The following list contains specific guidance to implementations concerning certain types of extensions.
An implementation must return exactly the number of return values specified by this standard unless the standard specifically indicates otherwise.
\itemitem{\b{Function behavior on non-standard data types}} An implementation does not define the behavior of \term{functions} on \term{types} not explicitly permitted by this standard.
No output can be produced by a function other than that specified in the standard or due to the signaling of conditions detected by the function.
Unsolicited output, such as garbage collection notifications and autoload heralds, should not go directly to the stream that is the value of a stream variable defined in this standard, but can go indirectly to terminal I/O by using a synonym stream to *terminal-io*.
Progress reports from such functions as load and compile are considered solicited, and are not covered by this prohibition.
Operators that are defined as \term{macros} or \term{special forms} may be defined as \term{functions} instead if the semantics can be preserved. or as follows: Alternate Proposal: MACRO-AS-FUNCTION:STATUS-QUO The standard will remain silent on the issue of whether or not is is valid for an implementation to implementation "macros" and "special forms" as functions.
Macros and special operators defined in this standard must not be functions.
The language described in this standard contains no subsets, though subsets are not forbidden.
For a language to be considered a subset, it must have the property that any valid program in that language has equivalent semantics and will run directly (with no extralingual pre-processing, and no special compatibility packages) in any conforming implementation of the full language.
A language that conforms to this requirement shall be described as being a “subset of Common Lisp as specified by ANSI ⟪standard number⟫.” Per X3J13. Fixed typo: "beginsubsection" => "beginsection".
Deprecated language features are not expected to appear in future Common Lisp standards, but are required to be implemented for conformance with this standard; see Section 1.5.1.1 (Required Language Features).
Conforming programs can use deprecated features; however, it is considered good programming style to avoid them. It is permissible for the compiler to produce style warnings about the use of such features at compile time, but there should be no such warnings at program execution time.
assoc-if-not | nsubst-if-not | require |
count-if-not | nsubstitute-if-not | set |
delete-if-not | position-if-not | subst-if-not |
find-if-not | provide | substitute-if-not |
gentemp | rassoc-if-not | |
member-if-not | remove-if-not |
Figure 1–2. Deprecated Functions
Restored PROVIDE,REQUIRE -kmp 7-Feb-92 Added GENTEMP -kmp 18-Aug-93
gensym has been deprecated.
The :test-not argument to the functions in the next figure are deprecated.
Figure 1–3. Functions with Deprecated :TEST-NOT Arguments
The use of the situation names compile, load, and eval in eval-when is deprecated.
*modules* is deprecated.
#S reader macro forces keyword names into the keyword package; see Section 2.4.8.13 (Sharpsign S). This feature is deprecated; in the future, keyword names will be taken in the package they are read in, so symbols that are actually in the keyword package should be used if that is what is desired.
The figures on the next twelve pages contain a complete enumeration of the 978 external symbols in the common-lisp package.
Figure 1–4. Symbols in the COMMON-LISP package (part one of twelve).
Figure 1–5. Symbols in the COMMON-LISP package (part two of twelve).
Figure 1–6. Symbols in the COMMON-LISP package (part three of twelve).
Figure 1–7. Symbols in the COMMON-LISP package (part four of twelve).
Figure 1–8. Symbols in the COMMON-LISP package (part five of twelve).
Figure 1–9. Symbols in the COMMON-LISP package (part six of twelve).
Figure 1–10. Symbols in the COMMON-LISP package (part seven of twelve).
Figure 1–11. Symbols in the COMMON-LISP package (part eight of twelve).
Figure 1–12. Symbols in the COMMON-LISP package (part nine of twelve).
Figure 1–13. Symbols in the COMMON-LISP package (part ten of twelve).
Figure 1–14. Symbols in the COMMON-LISP package (part eleven of twelve).
Figure 1–15. Symbols in the COMMON-LISP package (part twelve of twelve).