2 @c This is part of the GNU Emacs Lisp Reference Manual.
3 @c Copyright (C) 1990-1995, 1998-1999, 2001-2014 Free Software
5 @c See the file elisp.texi for copying conditions.
6 @node Control Structures
7 @chapter Control Structures
8 @cindex special forms for control structures
9 @cindex control structures
11 A Lisp program consists of a set of @dfn{expressions}, or
12 @dfn{forms} (@pxref{Forms}). We control the order of execution of
13 these forms by enclosing them in @dfn{control structures}. Control
14 structures are special forms which control when, whether, or how many
15 times to execute the forms they contain.
18 The simplest order of execution is sequential execution: first form
19 @var{a}, then form @var{b}, and so on. This is what happens when you
20 write several forms in succession in the body of a function, or at top
21 level in a file of Lisp code---the forms are executed in the order
22 written. We call this @dfn{textual order}. For example, if a function
23 body consists of two forms @var{a} and @var{b}, evaluation of the
24 function evaluates first @var{a} and then @var{b}. The result of
25 evaluating @var{b} becomes the value of the function.
27 Explicit control structures make possible an order of execution other
30 Emacs Lisp provides several kinds of control structure, including
31 other varieties of sequencing, conditionals, iteration, and (controlled)
32 jumps---all discussed below. The built-in control structures are
33 special forms since their subforms are not necessarily evaluated or not
34 evaluated sequentially. You can use macros to define your own control
35 structure constructs (@pxref{Macros}).
38 * Sequencing:: Evaluation in textual order.
39 * Conditionals:: @code{if}, @code{cond}, @code{when}, @code{unless}.
40 * Combining Conditions:: @code{and}, @code{or}, @code{not}.
41 * Iteration:: @code{while} loops.
42 * Nonlocal Exits:: Jumping out of a sequence.
48 @cindex sequential execution
50 Evaluating forms in the order they appear is the most common way
51 control passes from one form to another. In some contexts, such as in a
52 function body, this happens automatically. Elsewhere you must use a
53 control structure construct to do this: @code{progn}, the simplest
54 control construct of Lisp.
56 A @code{progn} special form looks like this:
60 (progn @var{a} @var{b} @var{c} @dots{})
65 and it says to execute the forms @var{a}, @var{b}, @var{c}, and so on, in
66 that order. These forms are called the @dfn{body} of the @code{progn} form.
67 The value of the last form in the body becomes the value of the entire
68 @code{progn}. @code{(progn)} returns @code{nil}.
70 @cindex implicit @code{progn}
71 In the early days of Lisp, @code{progn} was the only way to execute
72 two or more forms in succession and use the value of the last of them.
73 But programmers found they often needed to use a @code{progn} in the
74 body of a function, where (at that time) only one form was allowed. So
75 the body of a function was made into an ``implicit @code{progn}'':
76 several forms are allowed just as in the body of an actual @code{progn}.
77 Many other control structures likewise contain an implicit @code{progn}.
78 As a result, @code{progn} is not used as much as it was many years ago.
79 It is needed now most often inside an @code{unwind-protect}, @code{and},
80 @code{or}, or in the @var{then}-part of an @code{if}.
82 @defspec progn forms@dots{}
83 This special form evaluates all of the @var{forms}, in textual
84 order, returning the result of the final form.
88 (progn (print "The first form")
89 (print "The second form")
90 (print "The third form"))
91 @print{} "The first form"
92 @print{} "The second form"
93 @print{} "The third form"
94 @result{} "The third form"
99 Two other constructs likewise evaluate a series of forms but return
102 @defspec prog1 form1 forms@dots{}
103 This special form evaluates @var{form1} and all of the @var{forms}, in
104 textual order, returning the result of @var{form1}.
108 (prog1 (print "The first form")
109 (print "The second form")
110 (print "The third form"))
111 @print{} "The first form"
112 @print{} "The second form"
113 @print{} "The third form"
114 @result{} "The first form"
118 Here is a way to remove the first element from a list in the variable
119 @code{x}, then return the value of that former element:
122 (prog1 (car x) (setq x (cdr x)))
126 @defspec prog2 form1 form2 forms@dots{}
127 This special form evaluates @var{form1}, @var{form2}, and all of the
128 following @var{forms}, in textual order, returning the result of
133 (prog2 (print "The first form")
134 (print "The second form")
135 (print "The third form"))
136 @print{} "The first form"
137 @print{} "The second form"
138 @print{} "The third form"
139 @result{} "The second form"
145 @section Conditionals
146 @cindex conditional evaluation
148 Conditional control structures choose among alternatives. Emacs Lisp
149 has four conditional forms: @code{if}, which is much the same as in
150 other languages; @code{when} and @code{unless}, which are variants of
151 @code{if}; and @code{cond}, which is a generalized case statement.
153 @defspec if condition then-form else-forms@dots{}
154 @code{if} chooses between the @var{then-form} and the @var{else-forms}
155 based on the value of @var{condition}. If the evaluated @var{condition} is
156 non-@code{nil}, @var{then-form} is evaluated and the result returned.
157 Otherwise, the @var{else-forms} are evaluated in textual order, and the
158 value of the last one is returned. (The @var{else} part of @code{if} is
159 an example of an implicit @code{progn}. @xref{Sequencing}.)
161 If @var{condition} has the value @code{nil}, and no @var{else-forms} are
162 given, @code{if} returns @code{nil}.
164 @code{if} is a special form because the branch that is not selected is
165 never evaluated---it is ignored. Thus, in this example,
166 @code{true} is not printed because @code{print} is never called:
178 @defmac when condition then-forms@dots{}
179 This is a variant of @code{if} where there are no @var{else-forms},
180 and possibly several @var{then-forms}. In particular,
183 (when @var{condition} @var{a} @var{b} @var{c})
187 is entirely equivalent to
190 (if @var{condition} (progn @var{a} @var{b} @var{c}) nil)
194 @defmac unless condition forms@dots{}
195 This is a variant of @code{if} where there is no @var{then-form}:
198 (unless @var{condition} @var{a} @var{b} @var{c})
202 is entirely equivalent to
205 (if @var{condition} nil
206 @var{a} @var{b} @var{c})
210 @defspec cond clause@dots{}
211 @code{cond} chooses among an arbitrary number of alternatives. Each
212 @var{clause} in the @code{cond} must be a list. The @sc{car} of this
213 list is the @var{condition}; the remaining elements, if any, the
214 @var{body-forms}. Thus, a clause looks like this:
217 (@var{condition} @var{body-forms}@dots{})
220 @code{cond} tries the clauses in textual order, by evaluating the
221 @var{condition} of each clause. If the value of @var{condition} is
222 non-@code{nil}, the clause ``succeeds''; then @code{cond} evaluates its
223 @var{body-forms}, and returns the value of the last of @var{body-forms}.
224 Any remaining clauses are ignored.
226 If the value of @var{condition} is @code{nil}, the clause ``fails'', so
227 the @code{cond} moves on to the following clause, trying its @var{condition}.
229 A clause may also look like this:
236 Then, if @var{condition} is non-@code{nil} when tested, the @code{cond}
237 form returns the value of @var{condition}.
239 If every @var{condition} evaluates to @code{nil}, so that every clause
240 fails, @code{cond} returns @code{nil}.
242 The following example has four clauses, which test for the cases where
243 the value of @code{x} is a number, string, buffer and symbol,
248 (cond ((numberp x) x)
251 (setq temporary-hack x) ; @r{multiple body-forms}
252 (buffer-name x)) ; @r{in one clause}
253 ((symbolp x) (symbol-value x)))
257 Often we want to execute the last clause whenever none of the previous
258 clauses was successful. To do this, we use @code{t} as the
259 @var{condition} of the last clause, like this: @code{(t
260 @var{body-forms})}. The form @code{t} evaluates to @code{t}, which is
261 never @code{nil}, so this clause never fails, provided the @code{cond}
262 gets to it at all. For example:
267 (cond ((eq a 'hack) 'foo)
274 This @code{cond} expression returns @code{foo} if the value of @code{a}
275 is @code{hack}, and returns the string @code{"default"} otherwise.
278 Any conditional construct can be expressed with @code{cond} or with
279 @code{if}. Therefore, the choice between them is a matter of style.
284 (if @var{a} @var{b} @var{c})
286 (cond (@var{a} @var{b}) (t @var{c}))
291 * Pattern matching case statement::
294 @node Pattern matching case statement
295 @subsection Pattern matching case statement
297 @cindex pattern matching
299 To compare a particular value against various possible cases, the macro
300 @code{pcase} can come handy. It takes the following form:
303 (pcase @var{exp} @var{branch}1 @var{branch}2 @var{branch}3 @dots{})
306 where each @var{branch} takes the form @code{(@var{upattern}
307 @var{body-forms}@dots{})}.
309 It will first evaluate @var{exp} and then compare the value against each
310 @var{upattern} to see which @var{branch} to use, after which it will run the
311 corresponding @var{body-forms}. A common use case is to distinguish
312 between a few different constant values:
315 (pcase (get-return-code x)
316 (`success (message "Done!"))
317 (`would-block (message "Sorry, can't do it now"))
318 (`read-only (message "The shmliblick is read-only"))
319 (`access-denied (message "You do not have the needed rights"))
320 (code (message "Unknown return code %S" code)))
323 In the last clause, @code{code} is a variable that gets bound to the value that
324 was returned by @code{(get-return-code x)}.
326 To give a more complex example, a simple interpreter for a little
327 expression language could look like (note that this example requires
331 (defun evaluate (exp env)
333 (`(add ,x ,y) (+ (evaluate x env) (evaluate y env)))
334 (`(call ,fun ,arg) (funcall (evaluate fun env) (evaluate arg env)))
335 (`(fn ,arg ,body) (lambda (val)
336 (evaluate body (cons (cons arg val) env))))
338 ((pred symbolp) (cdr (assq exp env)))
339 (_ (error "Unknown expression %S" exp))))
342 Where @code{`(add ,x ,y)} is a pattern that checks that @code{exp} is a three
343 element list starting with the symbol @code{add}, then extracts the second and
344 third elements and binds them to the variables @code{x} and @code{y}.
345 @code{(pred numberp)} is a pattern that simply checks that @code{exp}
346 is a number, and @code{_} is the catch-all pattern that matches anything.
348 Here are some sample programs including their evaluation results:
351 (evaluate '(add 1 2) nil) ;=> 3
352 (evaluate '(add x y) '((x . 1) (y . 2))) ;=> 3
353 (evaluate '(call (fn x (add 1 x)) 2) nil) ;=> 3
354 (evaluate '(sub 1 2) nil) ;=> error
357 There are two kinds of patterns involved in @code{pcase}, called
358 @emph{U-patterns} and @emph{Q-patterns}. The @var{upattern} mentioned above
359 are U-patterns and can take the following forms:
362 @item `@var{qpattern}
363 This is one of the most common form of patterns. The intention is to mimic the
364 backquote macro: this pattern matches those values that could have been built
365 by such a backquote expression. Since we're pattern matching rather than
366 building a value, the unquote does not indicate where to plug an expression,
367 but instead it lets one specify a U-pattern that should match the value at
370 More specifically, a Q-pattern can take the following forms:
372 @item (@var{qpattern1} . @var{qpattern2})
373 This pattern matches any cons cell whose @code{car} matches @var{QPATTERN1} and
374 whose @code{cdr} matches @var{PATTERN2}.
375 @item [@var{qpattern1} @var{qpattern2} @dots{} @var{qpatternm}]
376 This pattern matches a vector of length @var{M} whose 0..(@var{M}-1)th
377 elements match @var{qpattern1}, @var{qpattern2} @dots{} @var{qpatternm},
380 This pattern matches any atom @code{equal} to @var{atom}.
381 @item ,@var{upattern}
382 This pattern matches any object that matches the @var{upattern}.
386 A mere symbol in a U-pattern matches anything, and additionally let-binds this
387 symbol to the value that it matched, so that you can later refer to it, either
388 in the @var{body-forms} or also later in the pattern.
390 This so-called @emph{don't care} pattern matches anything, like the previous
391 one, but unlike symbol patterns it does not bind any variable.
392 @item (pred @var{pred})
393 This pattern matches if the function @var{pred} returns non-@code{nil} when
394 called with the object being matched.
395 @item (or @var{upattern1} @var{upattern2}@dots{})
396 This pattern matches as soon as one of the argument patterns succeeds.
397 All argument patterns should let-bind the same variables.
398 @item (and @var{upattern1} @var{upattern2}@dots{})
399 This pattern matches only if all the argument patterns succeed.
400 @item (guard @var{exp})
401 This pattern ignores the object being examined and simply succeeds if @var{exp}
402 evaluates to non-@code{nil} and fails otherwise. It is typically used inside
403 an @code{and} pattern. For example, @code{(and x (guard (< x 10)))}
404 is a pattern which matches any number smaller than 10 and let-binds it to
405 the variable @code{x}.
408 @node Combining Conditions
409 @section Constructs for Combining Conditions
410 @cindex combining conditions
412 This section describes three constructs that are often used together
413 with @code{if} and @code{cond} to express complicated conditions. The
414 constructs @code{and} and @code{or} can also be used individually as
415 kinds of multiple conditional constructs.
418 This function tests for the falsehood of @var{condition}. It returns
419 @code{t} if @var{condition} is @code{nil}, and @code{nil} otherwise.
420 The function @code{not} is identical to @code{null}, and we recommend
421 using the name @code{null} if you are testing for an empty list.
424 @defspec and conditions@dots{}
425 The @code{and} special form tests whether all the @var{conditions} are
426 true. It works by evaluating the @var{conditions} one by one in the
429 If any of the @var{conditions} evaluates to @code{nil}, then the result
430 of the @code{and} must be @code{nil} regardless of the remaining
431 @var{conditions}; so @code{and} returns @code{nil} right away, ignoring
432 the remaining @var{conditions}.
434 If all the @var{conditions} turn out non-@code{nil}, then the value of
435 the last of them becomes the value of the @code{and} form. Just
436 @code{(and)}, with no @var{conditions}, returns @code{t}, appropriate
437 because all the @var{conditions} turned out non-@code{nil}. (Think
438 about it; which one did not?)
440 Here is an example. The first condition returns the integer 1, which is
441 not @code{nil}. Similarly, the second condition returns the integer 2,
442 which is not @code{nil}. The third condition is @code{nil}, so the
443 remaining condition is never evaluated.
447 (and (print 1) (print 2) nil (print 3))
454 Here is a more realistic example of using @code{and}:
458 (if (and (consp foo) (eq (car foo) 'x))
459 (message "foo is a list starting with x"))
464 Note that @code{(car foo)} is not executed if @code{(consp foo)} returns
465 @code{nil}, thus avoiding an error.
467 @code{and} expressions can also be written using either @code{if} or
468 @code{cond}. Here's how:
472 (and @var{arg1} @var{arg2} @var{arg3})
474 (if @var{arg1} (if @var{arg2} @var{arg3}))
476 (cond (@var{arg1} (cond (@var{arg2} @var{arg3}))))
481 @defspec or conditions@dots{}
482 The @code{or} special form tests whether at least one of the
483 @var{conditions} is true. It works by evaluating all the
484 @var{conditions} one by one in the order written.
486 If any of the @var{conditions} evaluates to a non-@code{nil} value, then
487 the result of the @code{or} must be non-@code{nil}; so @code{or} returns
488 right away, ignoring the remaining @var{conditions}. The value it
489 returns is the non-@code{nil} value of the condition just evaluated.
491 If all the @var{conditions} turn out @code{nil}, then the @code{or}
492 expression returns @code{nil}. Just @code{(or)}, with no
493 @var{conditions}, returns @code{nil}, appropriate because all the
494 @var{conditions} turned out @code{nil}. (Think about it; which one
497 For example, this expression tests whether @code{x} is either
498 @code{nil} or the integer zero:
501 (or (eq x nil) (eq x 0))
504 Like the @code{and} construct, @code{or} can be written in terms of
505 @code{cond}. For example:
509 (or @var{arg1} @var{arg2} @var{arg3})
517 You could almost write @code{or} in terms of @code{if}, but not quite:
521 (if @var{arg1} @var{arg1}
522 (if @var{arg2} @var{arg2}
528 This is not completely equivalent because it can evaluate @var{arg1} or
529 @var{arg2} twice. By contrast, @code{(or @var{arg1} @var{arg2}
530 @var{arg3})} never evaluates any argument more than once.
538 Iteration means executing part of a program repetitively. For
539 example, you might want to repeat some computation once for each element
540 of a list, or once for each integer from 0 to @var{n}. You can do this
541 in Emacs Lisp with the special form @code{while}:
543 @defspec while condition forms@dots{}
544 @code{while} first evaluates @var{condition}. If the result is
545 non-@code{nil}, it evaluates @var{forms} in textual order. Then it
546 reevaluates @var{condition}, and if the result is non-@code{nil}, it
547 evaluates @var{forms} again. This process repeats until @var{condition}
548 evaluates to @code{nil}.
550 There is no limit on the number of iterations that may occur. The loop
551 will continue until either @var{condition} evaluates to @code{nil} or
552 until an error or @code{throw} jumps out of it (@pxref{Nonlocal Exits}).
554 The value of a @code{while} form is always @code{nil}.
563 (princ (format "Iteration %d." num))
565 @print{} Iteration 0.
566 @print{} Iteration 1.
567 @print{} Iteration 2.
568 @print{} Iteration 3.
573 To write a ``repeat...until'' loop, which will execute something on each
574 iteration and then do the end-test, put the body followed by the
575 end-test in a @code{progn} as the first argument of @code{while}, as
582 (not (looking-at "^$"))))
587 This moves forward one line and continues moving by lines until it
588 reaches an empty line. It is peculiar in that the @code{while} has no
589 body, just the end test (which also does the real work of moving point).
592 The @code{dolist} and @code{dotimes} macros provide convenient ways to
593 write two common kinds of loops.
595 @defmac dolist (var list [result]) body@dots{}
596 This construct executes @var{body} once for each element of
597 @var{list}, binding the variable @var{var} locally to hold the current
598 element. Then it returns the value of evaluating @var{result}, or
599 @code{nil} if @var{result} is omitted. For example, here is how you
600 could use @code{dolist} to define the @code{reverse} function:
603 (defun reverse (list)
605 (dolist (elt list value)
606 (setq value (cons elt value)))))
610 @defmac dotimes (var count [result]) body@dots{}
611 This construct executes @var{body} once for each integer from 0
612 (inclusive) to @var{count} (exclusive), binding the variable @var{var}
613 to the integer for the current iteration. Then it returns the value
614 of evaluating @var{result}, or @code{nil} if @var{result} is omitted.
615 Here is an example of using @code{dotimes} to do something 100 times:
619 (insert "I will not obey absurd orders\n"))
624 @section Nonlocal Exits
625 @cindex nonlocal exits
627 A @dfn{nonlocal exit} is a transfer of control from one point in a
628 program to another remote point. Nonlocal exits can occur in Emacs Lisp
629 as a result of errors; you can also use them under explicit control.
630 Nonlocal exits unbind all variable bindings made by the constructs being
634 * Catch and Throw:: Nonlocal exits for the program's own purposes.
635 * Examples of Catch:: Showing how such nonlocal exits can be written.
636 * Errors:: How errors are signaled and handled.
637 * Cleanups:: Arranging to run a cleanup form if an error happens.
640 @node Catch and Throw
641 @subsection Explicit Nonlocal Exits: @code{catch} and @code{throw}
643 Most control constructs affect only the flow of control within the
644 construct itself. The function @code{throw} is the exception to this
645 rule of normal program execution: it performs a nonlocal exit on
646 request. (There are other exceptions, but they are for error handling
647 only.) @code{throw} is used inside a @code{catch}, and jumps back to
648 that @code{catch}. For example:
665 The @code{throw} form, if executed, transfers control straight back to
666 the corresponding @code{catch}, which returns immediately. The code
667 following the @code{throw} is not executed. The second argument of
668 @code{throw} is used as the return value of the @code{catch}.
670 The function @code{throw} finds the matching @code{catch} based on the
671 first argument: it searches for a @code{catch} whose first argument is
672 @code{eq} to the one specified in the @code{throw}. If there is more
673 than one applicable @code{catch}, the innermost one takes precedence.
674 Thus, in the above example, the @code{throw} specifies @code{foo}, and
675 the @code{catch} in @code{foo-outer} specifies the same symbol, so that
676 @code{catch} is the applicable one (assuming there is no other matching
677 @code{catch} in between).
679 Executing @code{throw} exits all Lisp constructs up to the matching
680 @code{catch}, including function calls. When binding constructs such
681 as @code{let} or function calls are exited in this way, the bindings
682 are unbound, just as they are when these constructs exit normally
683 (@pxref{Local Variables}). Likewise, @code{throw} restores the buffer
684 and position saved by @code{save-excursion} (@pxref{Excursions}), and
685 the narrowing status saved by @code{save-restriction}. It also runs
686 any cleanups established with the @code{unwind-protect} special form
687 when it exits that form (@pxref{Cleanups}).
689 The @code{throw} need not appear lexically within the @code{catch}
690 that it jumps to. It can equally well be called from another function
691 called within the @code{catch}. As long as the @code{throw} takes place
692 chronologically after entry to the @code{catch}, and chronologically
693 before exit from it, it has access to that @code{catch}. This is why
694 @code{throw} can be used in commands such as @code{exit-recursive-edit}
695 that throw back to the editor command loop (@pxref{Recursive Editing}).
697 @cindex CL note---only @code{throw} in Emacs
699 @b{Common Lisp note:} Most other versions of Lisp, including Common Lisp,
700 have several ways of transferring control nonsequentially: @code{return},
701 @code{return-from}, and @code{go}, for example. Emacs Lisp has only
702 @code{throw}. The @file{cl-lib} library provides versions of some of
703 these. @xref{Blocks and Exits,,,cl,Common Lisp Extensions}.
706 @defspec catch tag body@dots{}
707 @cindex tag on run time stack
708 @code{catch} establishes a return point for the @code{throw} function.
709 The return point is distinguished from other such return points by
710 @var{tag}, which may be any Lisp object except @code{nil}. The argument
711 @var{tag} is evaluated normally before the return point is established.
713 With the return point in effect, @code{catch} evaluates the forms of the
714 @var{body} in textual order. If the forms execute normally (without
715 error or nonlocal exit) the value of the last body form is returned from
718 If a @code{throw} is executed during the execution of @var{body},
719 specifying the same value @var{tag}, the @code{catch} form exits
720 immediately; the value it returns is whatever was specified as the
721 second argument of @code{throw}.
724 @defun throw tag value
725 The purpose of @code{throw} is to return from a return point previously
726 established with @code{catch}. The argument @var{tag} is used to choose
727 among the various existing return points; it must be @code{eq} to the value
728 specified in the @code{catch}. If multiple return points match @var{tag},
729 the innermost one is used.
731 The argument @var{value} is used as the value to return from that
735 If no return point is in effect with tag @var{tag}, then a @code{no-catch}
736 error is signaled with data @code{(@var{tag} @var{value})}.
739 @node Examples of Catch
740 @subsection Examples of @code{catch} and @code{throw}
742 One way to use @code{catch} and @code{throw} is to exit from a doubly
743 nested loop. (In most languages, this would be done with a ``goto''.)
744 Here we compute @code{(foo @var{i} @var{j})} for @var{i} and @var{j}
756 (throw 'loop (list i j)))
763 If @code{foo} ever returns non-@code{nil}, we stop immediately and return a
764 list of @var{i} and @var{j}. If @code{foo} always returns @code{nil}, the
765 @code{catch} returns normally, and the value is @code{nil}, since that
766 is the result of the @code{while}.
768 Here are two tricky examples, slightly different, showing two
769 return points at once. First, two return points with the same tag,
782 (print (catch2 'hack))
790 Since both return points have tags that match the @code{throw}, it goes to
791 the inner one, the one established in @code{catch2}. Therefore,
792 @code{catch2} returns normally with value @code{yes}, and this value is
793 printed. Finally the second body form in the outer @code{catch}, which is
794 @code{'no}, is evaluated and returned from the outer @code{catch}.
796 Now let's change the argument given to @code{catch2}:
801 (print (catch2 'quux))
808 We still have two return points, but this time only the outer one has
809 the tag @code{hack}; the inner one has the tag @code{quux} instead.
810 Therefore, @code{throw} makes the outer @code{catch} return the value
811 @code{yes}. The function @code{print} is never called, and the
812 body-form @code{'no} is never evaluated.
818 When Emacs Lisp attempts to evaluate a form that, for some reason,
819 cannot be evaluated, it @dfn{signals} an @dfn{error}.
821 When an error is signaled, Emacs's default reaction is to print an
822 error message and terminate execution of the current command. This is
823 the right thing to do in most cases, such as if you type @kbd{C-f} at
824 the end of the buffer.
826 In complicated programs, simple termination may not be what you want.
827 For example, the program may have made temporary changes in data
828 structures, or created temporary buffers that should be deleted before
829 the program is finished. In such cases, you would use
830 @code{unwind-protect} to establish @dfn{cleanup expressions} to be
831 evaluated in case of error. (@xref{Cleanups}.) Occasionally, you may
832 wish the program to continue execution despite an error in a subroutine.
833 In these cases, you would use @code{condition-case} to establish
834 @dfn{error handlers} to recover control in case of error.
836 Resist the temptation to use error handling to transfer control from
837 one part of the program to another; use @code{catch} and @code{throw}
838 instead. @xref{Catch and Throw}.
841 * Signaling Errors:: How to report an error.
842 * Processing of Errors:: What Emacs does when you report an error.
843 * Handling Errors:: How you can trap errors and continue execution.
844 * Error Symbols:: How errors are classified for trapping them.
847 @node Signaling Errors
848 @subsubsection How to Signal an Error
849 @cindex signaling errors
851 @dfn{Signaling} an error means beginning error processing. Error
852 processing normally aborts all or part of the running program and
853 returns to a point that is set up to handle the error
854 (@pxref{Processing of Errors}). Here we describe how to signal an
857 Most errors are signaled ``automatically'' within Lisp primitives
858 which you call for other purposes, such as if you try to take the
859 @sc{car} of an integer or move forward a character at the end of the
860 buffer. You can also signal errors explicitly with the functions
861 @code{error} and @code{signal}.
863 Quitting, which happens when the user types @kbd{C-g}, is not
864 considered an error, but it is handled almost like an error.
867 Every error specifies an error message, one way or another. The
868 message should state what is wrong (``File does not exist''), not how
869 things ought to be (``File must exist''). The convention in Emacs
870 Lisp is that error messages should start with a capital letter, but
871 should not end with any sort of punctuation.
873 @defun error format-string &rest args
874 This function signals an error with an error message constructed by
875 applying @code{format} (@pxref{Formatting Strings}) to
876 @var{format-string} and @var{args}.
878 These examples show typical uses of @code{error}:
882 (error "That is an error -- try something else")
883 @error{} That is an error -- try something else
887 (error "You have committed %d errors" 10)
888 @error{} You have committed 10 errors
892 @code{error} works by calling @code{signal} with two arguments: the
893 error symbol @code{error}, and a list containing the string returned by
896 @strong{Warning:} If you want to use your own string as an error message
897 verbatim, don't just write @code{(error @var{string})}. If @var{string}
898 contains @samp{%}, it will be interpreted as a format specifier, with
899 undesirable results. Instead, use @code{(error "%s" @var{string})}.
902 @defun signal error-symbol data
903 @anchor{Definition of signal}
904 This function signals an error named by @var{error-symbol}. The
905 argument @var{data} is a list of additional Lisp objects relevant to
906 the circumstances of the error.
908 The argument @var{error-symbol} must be an @dfn{error symbol}---a symbol
909 defined with @code{define-error}. This is how Emacs Lisp classifies different
910 sorts of errors. @xref{Error Symbols}, for a description of error symbols,
911 error conditions and condition names.
913 If the error is not handled, the two arguments are used in printing
914 the error message. Normally, this error message is provided by the
915 @code{error-message} property of @var{error-symbol}. If @var{data} is
916 non-@code{nil}, this is followed by a colon and a comma separated list
917 of the unevaluated elements of @var{data}. For @code{error}, the
918 error message is the @sc{car} of @var{data} (that must be a string).
919 Subcategories of @code{file-error} are handled specially.
921 The number and significance of the objects in @var{data} depends on
922 @var{error-symbol}. For example, with a @code{wrong-type-argument} error,
923 there should be two objects in the list: a predicate that describes the type
924 that was expected, and the object that failed to fit that type.
926 Both @var{error-symbol} and @var{data} are available to any error
927 handlers that handle the error: @code{condition-case} binds a local
928 variable to a list of the form @code{(@var{error-symbol} .@:
929 @var{data})} (@pxref{Handling Errors}).
931 The function @code{signal} never returns.
932 @c (though in older Emacs versions it sometimes could).
936 (signal 'wrong-number-of-arguments '(x y))
937 @error{} Wrong number of arguments: x, y
941 (signal 'no-such-error '("My unknown error condition"))
942 @error{} peculiar error: "My unknown error condition"
947 @cindex user errors, signaling
948 @defun user-error format-string &rest args
949 This function behaves exactly like @code{error}, except that it uses
950 the error symbol @code{user-error} rather than @code{error}. As the
951 name suggests, this is intended to report errors on the part of the
952 user, rather than errors in the code itself. For example,
953 if you try to use the command @code{Info-history-back} (@kbd{l}) to
954 move back beyond the start of your Info browsing history, Emacs
955 signals a @code{user-error}. Such errors do not cause entry to the
956 debugger, even when @code{debug-on-error} is non-@code{nil}.
957 @xref{Error Debugging}.
960 @cindex CL note---no continuable errors
962 @b{Common Lisp note:} Emacs Lisp has nothing like the Common Lisp
963 concept of continuable errors.
966 @node Processing of Errors
967 @subsubsection How Emacs Processes Errors
968 @cindex processing of errors
970 When an error is signaled, @code{signal} searches for an active
971 @dfn{handler} for the error. A handler is a sequence of Lisp
972 expressions designated to be executed if an error happens in part of the
973 Lisp program. If the error has an applicable handler, the handler is
974 executed, and control resumes following the handler. The handler
975 executes in the environment of the @code{condition-case} that
976 established it; all functions called within that @code{condition-case}
977 have already been exited, and the handler cannot return to them.
979 If there is no applicable handler for the error, it terminates the
980 current command and returns control to the editor command loop. (The
981 command loop has an implicit handler for all kinds of errors.) The
982 command loop's handler uses the error symbol and associated data to
983 print an error message. You can use the variable
984 @code{command-error-function} to control how this is done:
986 @defvar command-error-function
987 This variable, if non-@code{nil}, specifies a function to use to
988 handle errors that return control to the Emacs command loop. The
989 function should take three arguments: @var{data}, a list of the same
990 form that @code{condition-case} would bind to its variable;
991 @var{context}, a string describing the situation in which the error
992 occurred, or (more often) @code{nil}; and @var{caller}, the Lisp
993 function which called the primitive that signaled the error.
996 @cindex @code{debug-on-error} use
997 An error that has no explicit handler may call the Lisp debugger. The
998 debugger is enabled if the variable @code{debug-on-error} (@pxref{Error
999 Debugging}) is non-@code{nil}. Unlike error handlers, the debugger runs
1000 in the environment of the error, so that you can examine values of
1001 variables precisely as they were at the time of the error.
1003 @node Handling Errors
1004 @subsubsection Writing Code to Handle Errors
1005 @cindex error handler
1006 @cindex handling errors
1008 The usual effect of signaling an error is to terminate the command
1009 that is running and return immediately to the Emacs editor command loop.
1010 You can arrange to trap errors occurring in a part of your program by
1011 establishing an error handler, with the special form
1012 @code{condition-case}. A simple example looks like this:
1017 (delete-file filename)
1023 This deletes the file named @var{filename}, catching any error and
1024 returning @code{nil} if an error occurs. (You can use the macro
1025 @code{ignore-errors} for a simple case like this; see below.)
1027 The @code{condition-case} construct is often used to trap errors that
1028 are predictable, such as failure to open a file in a call to
1029 @code{insert-file-contents}. It is also used to trap errors that are
1030 totally unpredictable, such as when the program evaluates an expression
1033 The second argument of @code{condition-case} is called the
1034 @dfn{protected form}. (In the example above, the protected form is a
1035 call to @code{delete-file}.) The error handlers go into effect when
1036 this form begins execution and are deactivated when this form returns.
1037 They remain in effect for all the intervening time. In particular, they
1038 are in effect during the execution of functions called by this form, in
1039 their subroutines, and so on. This is a good thing, since, strictly
1040 speaking, errors can be signaled only by Lisp primitives (including
1041 @code{signal} and @code{error}) called by the protected form, not by the
1042 protected form itself.
1044 The arguments after the protected form are handlers. Each handler
1045 lists one or more @dfn{condition names} (which are symbols) to specify
1046 which errors it will handle. The error symbol specified when an error
1047 is signaled also defines a list of condition names. A handler applies
1048 to an error if they have any condition names in common. In the example
1049 above, there is one handler, and it specifies one condition name,
1050 @code{error}, which covers all errors.
1052 The search for an applicable handler checks all the established handlers
1053 starting with the most recently established one. Thus, if two nested
1054 @code{condition-case} forms offer to handle the same error, the inner of
1055 the two gets to handle it.
1057 If an error is handled by some @code{condition-case} form, this
1058 ordinarily prevents the debugger from being run, even if
1059 @code{debug-on-error} says this error should invoke the debugger.
1061 If you want to be able to debug errors that are caught by a
1062 @code{condition-case}, set the variable @code{debug-on-signal} to a
1063 non-@code{nil} value. You can also specify that a particular handler
1064 should let the debugger run first, by writing @code{debug} among the
1065 conditions, like this:
1070 (delete-file filename)
1071 ((debug error) nil))
1076 The effect of @code{debug} here is only to prevent
1077 @code{condition-case} from suppressing the call to the debugger. Any
1078 given error will invoke the debugger only if @code{debug-on-error} and
1079 the other usual filtering mechanisms say it should. @xref{Error Debugging}.
1081 @defmac condition-case-unless-debug var protected-form handlers@dots{}
1082 The macro @code{condition-case-unless-debug} provides another way to
1083 handle debugging of such forms. It behaves exactly like
1084 @code{condition-case}, unless the variable @code{debug-on-error} is
1085 non-@code{nil}, in which case it does not handle any errors at all.
1088 Once Emacs decides that a certain handler handles the error, it
1089 returns control to that handler. To do so, Emacs unbinds all variable
1090 bindings made by binding constructs that are being exited, and
1091 executes the cleanups of all @code{unwind-protect} forms that are
1092 being exited. Once control arrives at the handler, the body of the
1093 handler executes normally.
1095 After execution of the handler body, execution returns from the
1096 @code{condition-case} form. Because the protected form is exited
1097 completely before execution of the handler, the handler cannot resume
1098 execution at the point of the error, nor can it examine variable
1099 bindings that were made within the protected form. All it can do is
1100 clean up and proceed.
1102 Error signaling and handling have some resemblance to @code{throw} and
1103 @code{catch} (@pxref{Catch and Throw}), but they are entirely separate
1104 facilities. An error cannot be caught by a @code{catch}, and a
1105 @code{throw} cannot be handled by an error handler (though using
1106 @code{throw} when there is no suitable @code{catch} signals an error
1107 that can be handled).
1109 @defspec condition-case var protected-form handlers@dots{}
1110 This special form establishes the error handlers @var{handlers} around
1111 the execution of @var{protected-form}. If @var{protected-form} executes
1112 without error, the value it returns becomes the value of the
1113 @code{condition-case} form; in this case, the @code{condition-case} has
1114 no effect. The @code{condition-case} form makes a difference when an
1115 error occurs during @var{protected-form}.
1117 Each of the @var{handlers} is a list of the form @code{(@var{conditions}
1118 @var{body}@dots{})}. Here @var{conditions} is an error condition name
1119 to be handled, or a list of condition names (which can include @code{debug}
1120 to allow the debugger to run before the handler); @var{body} is one or more
1121 Lisp expressions to be executed when this handler handles an error.
1122 Here are examples of handlers:
1128 (arith-error (message "Division by zero"))
1130 ((arith-error file-error)
1132 "Either division by zero or failure to open a file"))
1136 Each error that occurs has an @dfn{error symbol} that describes what
1137 kind of error it is, and which describes also a list of condition names
1138 (@pxref{Error Symbols}). Emacs
1139 searches all the active @code{condition-case} forms for a handler that
1140 specifies one or more of these condition names; the innermost matching
1141 @code{condition-case} handles the error. Within this
1142 @code{condition-case}, the first applicable handler handles the error.
1144 After executing the body of the handler, the @code{condition-case}
1145 returns normally, using the value of the last form in the handler body
1146 as the overall value.
1148 @cindex error description
1149 The argument @var{var} is a variable. @code{condition-case} does not
1150 bind this variable when executing the @var{protected-form}, only when it
1151 handles an error. At that time, it binds @var{var} locally to an
1152 @dfn{error description}, which is a list giving the particulars of the
1153 error. The error description has the form @code{(@var{error-symbol}
1154 . @var{data})}. The handler can refer to this list to decide what to
1155 do. For example, if the error is for failure opening a file, the file
1156 name is the second element of @var{data}---the third element of the
1159 If @var{var} is @code{nil}, that means no variable is bound. Then the
1160 error symbol and associated data are not available to the handler.
1162 @cindex rethrow a signal
1163 Sometimes it is necessary to re-throw a signal caught by
1164 @code{condition-case}, for some outer-level handler to catch. Here's
1168 (signal (car err) (cdr err))
1172 where @code{err} is the error description variable, the first argument
1173 to @code{condition-case} whose error condition you want to re-throw.
1174 @xref{Definition of signal}.
1177 @defun error-message-string error-descriptor
1178 This function returns the error message string for a given error
1179 descriptor. It is useful if you want to handle an error by printing the
1180 usual error message for that error. @xref{Definition of signal}.
1183 @cindex @code{arith-error} example
1184 Here is an example of using @code{condition-case} to handle the error
1185 that results from dividing by zero. The handler displays the error
1186 message (but without a beep), then returns a very large number.
1190 (defun safe-divide (dividend divisor)
1192 ;; @r{Protected form.}
1193 (/ dividend divisor)
1197 (arith-error ; @r{Condition.}
1198 ;; @r{Display the usual message for this error.}
1199 (message "%s" (error-message-string err))
1201 @result{} safe-divide
1206 @print{} Arithmetic error: (arith-error)
1212 The handler specifies condition name @code{arith-error} so that it
1213 will handle only division-by-zero errors. Other kinds of errors will
1214 not be handled (by this @code{condition-case}). Thus:
1219 @error{} Wrong type argument: number-or-marker-p, nil
1223 Here is a @code{condition-case} that catches all kinds of errors,
1224 including those from @code{error}:
1236 ;; @r{This is a call to the function @code{error}.}
1237 (error "Rats! The variable %s was %s, not 35" 'baz baz))
1238 ;; @r{This is the handler; it is not a form.}
1239 (error (princ (format "The error was: %s" err))
1241 @print{} The error was: (error "Rats! The variable baz was 34, not 35")
1246 @defmac ignore-errors body@dots{}
1247 This construct executes @var{body}, ignoring any errors that occur
1248 during its execution. If the execution is without error,
1249 @code{ignore-errors} returns the value of the last form in @var{body};
1250 otherwise, it returns @code{nil}.
1252 Here's the example at the beginning of this subsection rewritten using
1253 @code{ignore-errors}:
1258 (delete-file filename))
1263 @defmac with-demoted-errors format body@dots{}
1264 This macro is like a milder version of @code{ignore-errors}. Rather
1265 than suppressing errors altogether, it converts them into messages.
1266 It uses the string @var{format} to format the message.
1267 @var{format} should contain a single @samp{%}-sequence; e.g.,
1268 @code{"Error: %S"}. Use @code{with-demoted-errors} around code
1269 that is not expected to signal errors, but
1270 should be robust if one does occur. Note that this macro uses
1271 @code{condition-case-unless-debug} rather than @code{condition-case}.
1275 @subsubsection Error Symbols and Condition Names
1276 @cindex error symbol
1278 @cindex condition name
1279 @cindex user-defined error
1280 @kindex error-conditions
1281 @kindex define-error
1283 When you signal an error, you specify an @dfn{error symbol} to specify
1284 the kind of error you have in mind. Each error has one and only one
1285 error symbol to categorize it. This is the finest classification of
1286 errors defined by the Emacs Lisp language.
1288 These narrow classifications are grouped into a hierarchy of wider
1289 classes called @dfn{error conditions}, identified by @dfn{condition
1290 names}. The narrowest such classes belong to the error symbols
1291 themselves: each error symbol is also a condition name. There are also
1292 condition names for more extensive classes, up to the condition name
1293 @code{error} which takes in all kinds of errors (but not @code{quit}).
1294 Thus, each error has one or more condition names: @code{error}, the
1295 error symbol if that is distinct from @code{error}, and perhaps some
1296 intermediate classifications.
1298 @defun define-error name message &optional parent
1299 In order for a symbol to be an error symbol, it must be defined with
1300 @code{define-error} which takes a parent condition (defaults to @code{error}).
1301 This parent defines the conditions that this kind of error belongs to.
1302 The transitive set of parents always includes the error symbol itself, and the
1303 symbol @code{error}. Because quitting is not considered an error, the set of
1304 parents of @code{quit} is just @code{(quit)}.
1307 @cindex peculiar error
1308 In addition to its parents, the error symbol has a @var{message} which
1309 is a string to be printed when that error is signaled but not handled. If that
1310 message is not valid, the error message @samp{peculiar error} is used.
1311 @xref{Definition of signal}.
1313 Internally, the set of parents is stored in the @code{error-conditions}
1314 property of the error symbol and the message is stored in the
1315 @code{error-message} property of the error symbol.
1317 Here is how we define a new error symbol, @code{new-error}:
1321 (define-error 'new-error "A new error" 'my-own-errors)
1326 This error has several condition names: @code{new-error}, the narrowest
1327 classification; @code{my-own-errors}, which we imagine is a wider
1328 classification; and all the conditions of @code{my-own-errors} which should
1329 include @code{error}, which is the widest of all.
1331 The error string should start with a capital letter but it should
1332 not end with a period. This is for consistency with the rest of Emacs.
1334 Naturally, Emacs will never signal @code{new-error} on its own; only
1335 an explicit call to @code{signal} (@pxref{Definition of signal}) in
1336 your code can do this:
1340 (signal 'new-error '(x y))
1341 @error{} A new error: x, y
1345 This error can be handled through any of its condition names.
1346 This example handles @code{new-error} and any other errors in the class
1347 @code{my-own-errors}:
1353 (my-own-errors nil))
1357 The significant way that errors are classified is by their condition
1358 names---the names used to match errors with handlers. An error symbol
1359 serves only as a convenient way to specify the intended error message
1360 and list of condition names. It would be cumbersome to give
1361 @code{signal} a list of condition names rather than one error symbol.
1363 By contrast, using only error symbols without condition names would
1364 seriously decrease the power of @code{condition-case}. Condition names
1365 make it possible to categorize errors at various levels of generality
1366 when you write an error handler. Using error symbols alone would
1367 eliminate all but the narrowest level of classification.
1369 @xref{Standard Errors}, for a list of the main error symbols
1370 and their conditions.
1373 @subsection Cleaning Up from Nonlocal Exits
1374 @cindex nonlocal exits, cleaning up
1376 The @code{unwind-protect} construct is essential whenever you
1377 temporarily put a data structure in an inconsistent state; it permits
1378 you to make the data consistent again in the event of an error or
1379 throw. (Another more specific cleanup construct that is used only for
1380 changes in buffer contents is the atomic change group; @ref{Atomic
1383 @defspec unwind-protect body-form cleanup-forms@dots{}
1384 @cindex cleanup forms
1385 @cindex protected forms
1386 @cindex error cleanup
1388 @code{unwind-protect} executes @var{body-form} with a guarantee that
1389 the @var{cleanup-forms} will be evaluated if control leaves
1390 @var{body-form}, no matter how that happens. @var{body-form} may
1391 complete normally, or execute a @code{throw} out of the
1392 @code{unwind-protect}, or cause an error; in all cases, the
1393 @var{cleanup-forms} will be evaluated.
1395 If @var{body-form} finishes normally, @code{unwind-protect} returns the
1396 value of @var{body-form}, after it evaluates the @var{cleanup-forms}.
1397 If @var{body-form} does not finish, @code{unwind-protect} does not
1398 return any value in the normal sense.
1400 Only @var{body-form} is protected by the @code{unwind-protect}. If any
1401 of the @var{cleanup-forms} themselves exits nonlocally (via a
1402 @code{throw} or an error), @code{unwind-protect} is @emph{not}
1403 guaranteed to evaluate the rest of them. If the failure of one of the
1404 @var{cleanup-forms} has the potential to cause trouble, then protect
1405 it with another @code{unwind-protect} around that form.
1407 The number of currently active @code{unwind-protect} forms counts,
1408 together with the number of local variable bindings, against the limit
1409 @code{max-specpdl-size} (@pxref{Definition of max-specpdl-size,, Local
1413 For example, here we make an invisible buffer for temporary use, and
1414 make sure to kill it before finishing:
1418 (let ((buffer (get-buffer-create " *temp*")))
1419 (with-current-buffer buffer
1422 (kill-buffer buffer))))
1427 You might think that we could just as well write @code{(kill-buffer
1428 (current-buffer))} and dispense with the variable @code{buffer}.
1429 However, the way shown above is safer, if @var{body-form} happens to
1430 get an error after switching to a different buffer! (Alternatively,
1431 you could write a @code{save-current-buffer} around @var{body-form},
1432 to ensure that the temporary buffer becomes current again in time to
1435 Emacs includes a standard macro called @code{with-temp-buffer} which
1436 expands into more or less the code shown above (@pxref{Definition of
1437 with-temp-buffer,, Current Buffer}). Several of the macros defined in
1438 this manual use @code{unwind-protect} in this way.
1441 Here is an actual example derived from an FTP package. It creates a
1442 process (@pxref{Processes}) to try to establish a connection to a remote
1443 machine. As the function @code{ftp-login} is highly susceptible to
1444 numerous problems that the writer of the function cannot anticipate, it
1445 is protected with a form that guarantees deletion of the process in the
1446 event of failure. Otherwise, Emacs might fill up with useless
1454 (setq process (ftp-setup-buffer host file))
1455 (if (setq win (ftp-login process host user password))
1456 (message "Logged in")
1457 (error "Ftp login failed")))
1458 (or win (and process (delete-process process)))))
1462 This example has a small bug: if the user types @kbd{C-g} to
1463 quit, and the quit happens immediately after the function
1464 @code{ftp-setup-buffer} returns but before the variable @code{process} is
1465 set, the process will not be killed. There is no easy way to fix this bug,
1466 but at least it is very unlikely.