24.2 Debugging Tools

The utilities described in this section are sufficiently complex and sufficiently dependent on the host environment that their complete definition is beyond the scope of this book. However, they are also sufficiently useful to warrant mention here. It is expected that every implementation will provide some version of these utilities, however clever or however simple.

Коммунальные услуги, описанные в этом разделе достаточно сложны и достаточно зависит от внешней среды, что их полное описание выходит за рамки этой книги. Тем не менее, они также достаточно полезно, чтобы оправдать упомянуть здесь. Ожидается, что каждая реализация предоставить некоторые версии этих программ, однако умный или же просто.

Описанные в этом разделе утилиты достаточно сложны и зависят от внешней среды ОС, что их полное описание выходит за рамки книги. Тем не менее их описание будет полезным. Предполагается, что каждая реализация будет представлять некоторую версию этих утилит.

[Макрос] trace { function-name}*

[Макрос] untrace { function-name}*

Invoking trace with one or more function-names (symbols or lists, whose car is setf—see section 7.1), causes the functions named to be traced. Henceforth, whenever such a function is invoked, information about the call, the arguments passed, and the eventually returned values, if any, will be printed to the stream that is the value of *trace-output*. For example:

(trace fft gcd string-upcase)

If a function call is open-coded (possibly as a result of an inline declaration), then such a call may not produce trace output.

Invoking untrace with one or more function names will cause those functions not to be traced any more.

Tracing an already traced function, or untracing a function not currently being traced, should produce no harmful effects but may produce a warning message.

Calling trace with no argument forms will return a list of functions currently being traced.

Calling untrace with no argument forms will cause all currently traced functions to be no longer traced.

The values returned by trace and untrace when given argument forms are implementation-dependent.

trace and untrace may also accept additional implementation-dependent argument formats. The format of the trace output is implementation-dependent.


[Макрос] step form

This evaluates form and returns what form returns. However, the user is allowed to interactively “single-step” through the evaluation of form, at least through those evaluation steps that are performed interpretively. The nature of the interaction is implementation-dependent. However, implementations are encouraged to respond to the typing of the character ? by providing help, including a list of commands.

step evaluates its argument form in the current lexical environment (not simply a null environment), and that calls to step may be compiled, in which case an implementation may step through only those parts of the evaluation that are interpreted. (In other words, the form itself is unlikely to be stepped, but if executing it happens to invoke interpreted code, then that code may be stepped.)


[Макрос] time form

This evaluates form and returns what form returns. However, as a side effect, various timing data and other information are printed to the stream that is the value of *trace-output*. The nature and format of the printed information is implementation-dependent. However, implementations are encouraged to provide such information as elapsed real time, machine run time, storage management statistics, and so on.

time evaluates its argument form in the current lexical environment (not simply a null environment), and that calls to time may be compiled.


[Function] describe object &optional stream

describe prints, to the stream information about the object. Sometimes it will describe something that it finds inside something else; such recursive descriptions are indented appropriately. For instance, describe of a symbol will exhibit the symbol’s value, its definition, and each of its properties. describe of a floating-point number will exhibit its internal representation in a way that is useful for tracking down round-off errors and the like. The nature and format of the output is implementation-dependent.

describe returns no values (that is, it returns what the expression (values) returns: zero values).

The output is sent to the specified stream, which defaults to the value of *standard-output*; the stream may also be nil (meaning *standard-output*) or t (meaning *terminal-io*).

The behavior of describe depends on the generic function describe-object (see below).


That describe is forbidden to prompt for or require user input when given exactly one argument; It is permitted implementations to extend describe to accept keyword arguments that may cause it to prompt for or to require user input.

[Generic function] describe-object object stream
[Primary method] describe-object (object standard-object) stream

The generic function describe-object writes a description of an object to a stream. The function describe-object is called by the describe function; it should not be called by the user.

Each implementation must provide a method on the class standard-object and methods on enough other classes to ensure that there is always an applicable method. Implementations are free to add methods for other classes. Users can write methods for describe-object for their own classes if they do not wish to inherit an implementation-supplied method.

The first argument may be any Lisp object. The second argument is a stream; it cannot be t or nil. The values returned by describe-object are unspecified.

Methods on describe-object may recursively call describe. Indentation, depth limits, and circularity detection are all taken care of automatically, provided that each method handles exactly one level of structure and calls describe recursively if there are more structural levels. If this rule is not obeyed, the results are undefined.

In some implementations the stream argument passed to a describe-object method is not the original stream but is an intermediate stream that implements parts of describe. Methods should therefore not depend on the identity of this stream. ___________________________________________________________________

Обоснование: This proposal was closely modeled on the CLOS description of print-object, which was well thought out and provides a great deal of functionality and implementation freedom. Implementation techniques for print-object are applicable to describe-object.

The reason for making the return values for describe-object unspecified is to avoid forcing users to write (values) explicitly in all their methods; describe should take care of that.

___________________________________________________________________________________________________________

[Function] inspect object

inspect is an interactive version of describe. The nature of the interaction is implementation-dependent, but the purpose of inspect is to make it easy to wander through a data structure, examining and modifying parts of it. Implementations are encouraged to respond to the typing of the character ? by providing help, including a list of commands.

The values returned by inspect are implementation-dependent.


[Function] room &optional x

room prints, to the stream in the variable *standard-output*, information about the state of internal storage and its management. This might include descriptions of the amount of memory in use and the degree of memory compaction, possibly broken down by internal data type if that is appropriate. The nature and format of the printed information is implementation-dependent. The intent is to provide information that may help a user to tune a program to a particular implementation.

(room nil) prints out a minimal amount of information. (room t) prints out a maximal amount of information. Simply (room) prints out an intermediate amount of information that is likely to be useful.

The argument x may also be the keyword :default, which has the same effect as passing no argument at all.


[Function] ed &optional x

If the implementation provides a resident editor, this function should invoke it.

(ed) or (ed nil) simply enters the editor, leaving you in the same state as the last time you were in the editor.

(ed pathname) edits the contents of the file specified by pathname. The pathname may be an actual pathname or a string.

ed accepts logical pathnames (see section 23.1.5).

(ed symbol) tries to let you edit the text for the function named symbol. The means by which the function text is obtained is implementation-dependent; it might involve searching the file system, or pretty printing resident interpreted code, for example.

Function name may be any function-name (a symbol or a list whose car is setf—see section 7.1). Thus one may write (ed ’(setf cadr)) to edit the setf expansion function for cadr.


[Function] dribble &optional pathname

(dribble pathname) may rebind *standard-input* and *standard-output*, and may take other appropriate action, so as to send a record of the input/output interaction to a file named by pathname. The primary purpose of this is to create a readable record of an interactive session.

(dribble) terminates the recording of input and output and closes the dribble file.

dribble also accepts logical pathnames (see section 23.1.5).

dribble is intended primarily for interactive debugging and that its effect cannot be relied upon for use in portable programs.

Different implementations of Common Lisp have used radically different techniques for implementing dribble. All are reasonable interpretations of the original specification, and all behave in approximately the same way if dribble is called only from the interactive top level. However, they may have quite different behaviors if dribble is called from within compound forms.

Consider two models of the operation of dribble. In the “redirecting” model, a call to dribble with a pathname argument alters certain global variables such as *standard-output*, perhaps by constructing a broadcast stream directed to both the original value of *standard-output* and to the dribble file; other streams may be affected as well. A call to dribble with no arguments undoes these side effects.

In the “recursive” model, by contrast, a call to dribble with a pathname argument creates a new interactive command loop and calls it recursively. This new command loop is just like an ordinary read-eval-print loop except that it also echoes the interaction to the dribble file. A call to dribble with no arguments does a throw that exits the recursive command loop and returns to the original caller of dribble with an argument.

The two models may be distinguished by this test case:

(progn (dribble "basketball")
       (print "Larry")
       (dribble)
       (princ "Bird"))

If this form is input to the Lisp top level, in either model a newline (provided by the function print) and the words Larry Bird will be printed to the standard output. The redirecting dribble model will additionally print all but the word Bird to a file named basketball.

By contrast, the recursive dribble model will enter a recursive command loop and not print anything until (dribble) is executed from within the new interactive command loop. At that time the file named basketball will be closed, and then execution of the progn form will be resumed. A newline and “Larry ” (note the trailing space) will be printed to the standard output, and then the call (dribble) may complain that there is no active dribble file. Once this error is resolved, the word Bird may be printed to the standard output.

Here is a slightly different test case:

(dribble "baby-food")

(progn (print "Mashed banana")
       (dribble)
       (princ "and cream of rice"))

If this form is input to the Lisp top level, in the redirecting model a newline and the words Mashed banana and cream of rice will be printed to the standard output and all but the words and cream of rice will be sent to a file named baby-food.

The recursive model will direct exactly the same output to the file named baby-food but will never print the words and cream of rice to the standard output because the call (dribble) does not return normally; it throws.

The redirecting model may be intuitively more appealing to some. The recursive model, however, may be more robust; it carefully limits the extent of the dribble operation and disables dribbling if a throw of any kind occurs. The vote by X3J13 was an explicit decision not to decide which model to use. Users are advised to call dribble only interactively, at top level.


[Function] apropos string &optional package
[Function] apropos-list string &optional package

(apropos string) tries to find all available symbols whose print names contain string as a substring. (A symbol may be supplied for the string, in which case the print name of the symbol is used.) Whenever apropos finds a symbol, it prints out the symbol’s name; in addition, information about the function definition and dynamic value of the symbol, if any, is printed. If package is specified and not nil, then only symbols available in that package are examined; otherwise “all” packages are searched, as if by do-all-symbols. Because a symbol may be available by way of more than one inheritance path, apropos may print information about the same symbol more than once. The information is printed to the stream that is the value of *standard-output*. apropos returns no values (that is, it returns what the expression (values) returns: zero values).

apropos-list performs the same search that apropos does but prints nothing. It returns a list of the symbols whose print names contain string as a substring.