guile/doc/ref/scheme-debugging.texi

@c -*-texinfo-*-
@c This is part of the GNU Guile Reference Manual.
@c Copyright (C)  1996, 1997, 2000, 2001, 2002, 2003, 2004
@c   Free Software Foundation, Inc.
@c See the file guile.texi for copying conditions.

@page
@node Debugging Features
@section Debugging Features

Guile includes debugging tools to help you work out what is going wrong
when a program signals an error or behaves differently to how you would
expect.  This chapter describes how to use these tools.

Broadly speaking, Guile's debugging support allows you to do two things:

@itemize @bullet
@item
specify @dfn{breakpoints} --- points in the execution of a program where
execution should pause so you can see what is going on

@item
examine in detail the ``scene of the crime'' --- in other words, the
execution context at a breakpoint, or when the last error occurred.
@end itemize

@noindent
The details are more complex and more powerful @dots{}

@menu
* Examples::
* Intro to Breakpoints::        Setting and manipulating them.
* Tracing::                     Tracing program execution.
@end menu


@node Examples
@subsection Examples

Before we dive into the details and reference documentation of
guile-debugging's features, this chapter sets the scene by presenting a
few examples of what you can do with guile-debugging.

@menu
* Single Stepping through a Procedure's Code::
* Profiling or Tracing a Procedure's Code::
@end menu


@node Single Stepping through a Procedure's Code
@subsubsection Single Stepping through a Procedure's Code

A good way to explore in detail what a Scheme procedure does is to set a
breakpoint on it and then single step through what it does.  To do this,
use the @code{break-in} procedure from the @code{(ice-9 debugging
breakpoints)} module with the @code{debug-trap} behaviour from
@code{(ice-9 debugging ice-9-debugger-extensions)}.  The following
sample session illustrates this.  It assumes that the file
@file{matrix.scm} defines a procedure @code{mkmatrix}, which is the one
we want to explore, and another procedure @code{do-main} which calls
@code{mkmatrix}.

@lisp
$ /usr/bin/guile -q
guile> (use-modules (ice-9 debugger)
                    (ice-9 debugging ice-9-debugger-extensions)
                    (ice-9 debugging breakpoints))
guile> (load "matrix.scm")
guile> (break-in 'mkmatrix #:behaviour debug-trap)
#<<break-in> 808cb70>
guile> (do-main 4)
This is the Guile debugger -- for help, type `help'.
There are 3 frames on the stack.

Frame 2 at matrix.scm:8:3
        [mkmatrix]
debug> next
Frame 3 at matrix.scm:4:3
        (let ((x 1)) (quote this-is-a-matric))
debug> info frame
Stack frame: 3
This frame is an evaluation.
The expression being evaluated is:
matrix.scm:4:3:
  (let ((x 1)) (quote this-is-a-matric))
debug> next
Frame 3 at matrix.scm:5:21
        (quote this-is-a-matric)
debug> bt
In unknown file:
   ?: 0* [primitive-eval (do-main 4)]
In standard input:
   4: 1* [do-main 4]
In matrix.scm:
   8: 2  [mkmatrix]
   ...
   5: 3  (quote this-is-a-matric)
debug> quit
this-is-a-matric
guile> 
@end lisp

Or you can use guile-debugging's Emacs interface (GDS), by using the
module @code{(ice-9 gds-client)} instead of @code{(ice-9 debugger)} and
@code{(ice-9 debugging ice-9-debugger-extensions)}, and changing
@code{debug-trap} to @code{gds-debug-trap}.  Then the stack and
corresponding source locations are displayed in Emacs instead of on the
Guile command line.


@node Profiling or Tracing a Procedure's Code
@subsubsection Profiling or Tracing a Procedure's Code

What if you wanted to get a trace of everything that the Guile evaluator
does within a given procedure, but without Guile stopping and waiting
for your input at every step?  In this case you set a breakpoint on the
procedure using @code{break-in} (the same as in the previous example),
but use the @code{trace-trap} and @code{trace-until-exit} behaviours
provided by the @code{(ice-9 debugging trace)} module.

@lisp
guile> (use-modules (ice-9 debugging breakpoints) (ice-9 debugging trace))
guile> (load "matrix.scm")
guile> (break-in 'mkmatrix #:behaviour (list trace-trap trace-until-exit))
#<<break-in> 808b430>
guile> (do-main 4)
|  2: [mkmatrix]
|  3: [define (define yy 23) ((()) #<eval-closure 4028db30>)]
|  3: [define (define yy 23) ((()) #<eval-closure 4028db30>)]
|  3: =>(#@@define yy 23)
|  3: [let (let # #) (# #)]
|  3: [let (let # #) (# #)]
|  3: =>(#@@let* (x 1) #@@let (quote this-is-a-matric))
|  2: (letrec ((yy 23)) (let ((x 1)) (quote this-is-a-matric)))
|  3: [let (let # #) (# # #)]
|  3: [let (let # #) (# # #)]
|  3: =>(#@@let* (x 1) #@@let (quote this-is-a-matric))
|  2: (let ((x 1)) (quote this-is-a-matric))
|  3: [quote (quote this-is-a-matric) ((x . 1) ((yy) 23) (()) ...)]
|  3: [quote (quote this-is-a-matric) ((x . 1) ((yy) 23) (()) ...)]
|  3: =>(#@@quote this-is-a-matric)
|  2: (quote this-is-a-matric)
|  2: =>this-is-a-matric
this-is-a-matric
guile> (do-main 4)
|  2: [mkmatrix]
|  2: (letrec ((yy 23)) (let ((x 1)) (quote this-is-a-matric)))
|  2: (let ((x 1)) (quote this-is-a-matric))
|  2: (quote this-is-a-matric)
|  2: =>this-is-a-matric
this-is-a-matric
guile> 
@end lisp

This example shows the default configuration for how each line of trace
output is formatted, which is:

@itemize
@item
the character @code{|}, a visual clue that the line is a line of trace
output, followed by

@item
a number indicating the real evaluator stack depth (where ``real'' means
not counting tail-calls), followed by

@item
a summary of the expression being evaluated (@code{(@dots{})}), the
procedure being called (@code{[@dots{}]}), or the value being returned
from an evaluation or procedure call (@code{=>@dots{}}).
@end itemize

@noindent
You can customize @code{(ice-9 debugging trace)} to show different
information in each trace line using the @code{set-trace-layout}
procedure.  The next example shows how to get the source location in
each trace line instead of the stack depth.

@lisp
guile> (set-trace-layout "|~16@@a: ~a\n" trace/source trace/info)
guile> (do-main 4)
|  matrix.scm:7:2: [mkmatrix]
|                : (letrec ((yy 23)) (let ((x 1)) (quote this-is-a-matric)))
|  matrix.scm:3:2: (let ((x 1)) (quote this-is-a-matric))
| matrix.scm:4:20: (quote this-is-a-matric)
| matrix.scm:4:20: =>this-is-a-matric
this-is-a-matric
guile> 
@end lisp

(For anyone wondering why the first @code{(do-main 4)} call above
generates lots more trace lines than the subsequent calls: these
examples also demonstrate how the Guile evaluator ``memoizes'' code.
When Guile evaluates a source code expression for the first time, it
changes some parts of the expression so that they will be quicker to
evaluate when that expression is evaluated again; this is called
memoization.  The trace output from the first @code{(do-main 4)} call
shows memoization steps, such as an internal define being transformed to
a letrec.)


@node Intro to Breakpoints
@subsection Intro to Breakpoints

Sometimes a piece of Scheme code isn't working and you'd like to go
through it step by step.  You can do this in Guile by setting a
breakpoint at the start of the relevant code, and then using the command
line or Emacs interface to step through it.

A breakpoint can be specified by procedure name or by location -- the
relevant code's file name, line number and column number.  For details
please see the full documentation for @code{break-in} and
@code{break-at} in @ref{Intro to Breakpoints}.

When you set a breakpoint, you can specify any ``behaviour'' you like
for what should happen when the breakpoint is hit; a breakpoint
``behaviour'' is just a Scheme procedure with the right signature.

@menu
* Breakpoints Overview::
@end menu


@node Breakpoints Overview
@subsubsection How Breakpoints Work and Why They Are Useful

Often, debugging the last error is not enough to tell you what went
wrong.  For example, the root cause of the error may have arisen a long
time before the error was signalled, in which case the execution context
of the error is too late to be useful.  Or your program might not signal
an error at all, just return an unexpected result or have some incorrect
side effect.

In many such cases, it's useful to pause the program at or before the
point where you suspect the problem arises.  Then you can explore the
stack, display the values of key variables, and generally check that the
state of the program is as you expect.  If all is well, you can let the
program continue running normally, or step more slowly through each
expression that the Scheme interpreter evaluates.  Single-stepping may
reveal that the program is going through blocks of code that you didn't
intend --- a useful data point for understanding what the underlying
problem is.

Telling Guile where or when to pause a program is called @dfn{setting a
breakpoint}.  When a breakpoint is hit, Guile's default behaviour is to
enter the interactive debugger, where there are now two sets of commands
available:

@itemize @bullet
@item
all the commands as described for last error debugging
(@pxref{Interactive Debugger}), which allow you to explore the stack and
so on

@item
additional commands for continuing program execution in various ways:
@code{next}, @code{step}, @code{finish}, @code{trace-finish} and
@code{continue}.
@end itemize

Use of the interactive debugger is described in @ref{Interactive
Debugger}.


@node Tracing
@subsection Tracing

Tracing has already been described as a breakpoint behaviour, but we
mention it again here because it is so useful, and because Guile
actually now has @emph{two} mechanisms for tracing, and its worth
clarifying the differences between them.

@menu
* Old Tracing::                 Tracing provided by (ice-9 debug).
* New Tracing::                 Breakpoint-based tracing.
* Tracing Compared::            Differences between old and new.
@end menu


@node Old Tracing
@subsubsection Tracing Provided by @code{(ice-9 debug)}

The @code{(ice-9 debug)} module implements tracing of procedure
applications.  When a procedure is @dfn{traced}, it means that every
call to that procedure is reported to the user during a program run.
The idea is that you can mark a collection of procedures for tracing,
and Guile will subsequently print out a line of the form

@smalllisp
|  |  [@var{procedure} @var{args} @dots{}]
@end smalllisp

whenever a marked procedure is about to be applied to its arguments.
This can help a programmer determine whether a function is being called
at the wrong time or with the wrong set of arguments.

In addition, the indentation of the output is useful for demonstrating
how the traced applications are or are not tail recursive with respect
to each other.  Thus, a trace of a non-tail recursive factorial
implementation looks like this:

@smalllisp
[fact1 4]
|  [fact1 3]
|  |  [fact1 2]
|  |  |  [fact1 1]
|  |  |  |  [fact1 0]
|  |  |  |  1
|  |  |  1
|  |  2
|  6
24
@end smalllisp

While a typical tail recursive implementation would look more like this:

@smalllisp
[fact2 4]
[facti 1 4]
[facti 4 3]
[facti 12 2]
[facti 24 1]
[facti 24 0]
24
@end smalllisp

@deffn {Scheme Procedure} trace procedure
Enable tracing for @code{procedure}.  While a program is being run,
Guile will print a brief report at each call to a traced procedure,
advising the user which procedure was called and the arguments that were
passed to it.
@end deffn

@deffn {Scheme Procedure} untrace procedure
Disable tracing for @code{procedure}.
@end deffn

Here is another example:

@lisp
(define (rev ls)
  (if (null? ls)
      '()
      (append (rev (cdr ls))
              (cons (car ls) '())))) @result{} rev

(trace rev) @result{} (rev)

(rev '(a b c d e))
@result{} [rev (a b c d e)]
   |  [rev (b c d e)]
   |  |  [rev (c d e)]
   |  |  |  [rev (d e)]
   |  |  |  |  [rev (e)]
   |  |  |  |  |  [rev ()]
   |  |  |  |  |  ()
   |  |  |  |  (e)
   |  |  |  (e d)
   |  |  (e d c)
   |  (e d c b)
   (e d c b a)
   (e d c b a)
@end lisp

Note the way Guile indents the output, illustrating the depth of
execution at each procedure call.  This can be used to demonstrate, for
example, that Guile implements self-tail-recursion properly:
 
@lisp
(define (rev ls sl)
  (if (null? ls)
      sl
      (rev (cdr ls)
           (cons (car ls) sl)))) @result{} rev
 
(trace rev) @result{} (rev)
 
(rev '(a b c d e) '())
@result{} [rev (a b c d e) ()]
   [rev (b c d e) (a)]
   [rev (c d e) (b a)]
   [rev (d e) (c b a)]
   [rev (e) (d c b a)]
   [rev () (e d c b a)]
   (e d c b a)
   (e d c b a)
@end lisp
 
Since the tail call is effectively optimized to a @code{goto} statement,
there is no need for Guile to create a new stack frame for each
iteration.  Tracing reveals this optimization in operation.


@node New Tracing
@subsubsection Breakpoint-based Tracing

Guile's newer mechanism implements tracing as an optional behaviour for
any kind of breakpoint.

To trace a procedure (in the same kind of way as the older tracing), use
the @code{trace!} procedure to set a procedure breakpoint with
@code{trace-here} behaviour:

@lisp
(trace! fact1)
@print{}
Set breakpoint 1: [fact1]
@result{}
#<<procedure-breakpoint> 40337bf0>

(fact1 4)
@print{}
|  [fact1 4]
|  |  [fact1 3]
|  |  |  [fact1 2]
|  |  |  |  [fact1 1]
|  |  |  |  |  [fact1 0]
|  |  |  |  |  1
|  |  |  |  2
|  |  |  6
|  |  24
|  24
@result{}
24
@end lisp

To trace evaluation of a source expression, evaluate code containing a
breakpoint marker @code{##} in the appropriate place, then use
@code{set-breakpoint} to change the behaviour of the new breakpoint to
@code{trace-here}:

@lisp
(define (fact1 n)
  (if ##(= n 0)
      1
      (* n (fact1 (- n 1)))))
@print{}
Set breakpoint 4: standard input:13:9: (= n 0)

(use-modules (ice-9 debugger behaviour))
(set-breakpoint! trace-here 4)
@print{}
Breakpoint 4: standard input:13:9: (= n 0)
        enabled? = #t
        behaviour = #<procedure trace-here ()>

(fact1 4)
@print{}
|  (= n 0)
|  #f
|  (= n 0)
|  #f
|  (= n 0)
|  #f
|  (= n 0)
|  #f
|  (= n 0)
|  #t
@result{}
24
@end lisp

@noindent
(Note --- this example reveals a bug: each occurrence of @code{(= n 0)}
should be shown indented with respect to the one before it, as
@code{fact1} does not call itself tail-recursively.)

You can also give a breakpoint the @code{trace-subtree} behaviour, which
means to trace the breakpoint location itself plus any evaluations and
applications that occur below it in the call stack.  In the following
example, this allows us to see the evaluated arguments that are being
compared by the @code{=} procedure:

@lisp
(set-breakpoint! trace-subtree 4)
@print{}
Breakpoint 4: standard input:13:9: (= n 0)
        enabled? = #t
        behaviour = #<procedure trace-subtree ()>

(fact1 4)
@print{}
|  (= n 0)
|  [= 4 0]
|  #f
|  (= n 0)
|  [= 3 0]
|  #f
|  (= n 0)
|  [= 2 0]
|  #f
|  (= n 0)
|  [= 1 0]
|  #f
|  (= n 0)
|  [= 0 0]
|  #t
@result{}
24
@end lisp


@node Tracing Compared
@subsubsection Differences Between Old and New Tracing Mechanisms

The newer tracing mechanism is more general and so more powerful than
the older one: it works for expressions as well as procedure
applications, and it implements the useful @code{trace-subtree}
behaviour as well as the more traditional @code{trace-here}.

The older mechanism will probably become obsolete eventually, but it's
worth keeping it around for a while until we are sure that the new
mechanism is correct and does what programmers need.