3 perldebug - Perl debugging
7 First of all, have you tried using the B<-w> switch?
9 =head1 The Perl Debugger
11 "As soon as we started programming, we found to our
12 surprise that it wasn't as easy to get programs right
13 as we had thought. Debugging had to be discovered.
14 I can remember the exact instant when I realized that
15 a large part of my life from then on was going to be
16 spent in finding mistakes in my own programs."
18 I< --Maurice Wilkes, 1949>
20 If you invoke Perl with the B<-d> switch, your script runs under the
21 Perl source debugger. This works like an interactive Perl
22 environment, prompting for debugger commands that let you examine
23 source code, set breakpoints, get stack backtraces, change the values of
24 variables, etc. This is so convenient that you often fire up
25 the debugger all by itself just to test out Perl constructs
26 interactively to see what they do. For example:
30 In Perl, the debugger is not a separate program as it usually is in the
31 typical compiled environment. Instead, the B<-d> flag tells the compiler
32 to insert source information into the parse trees it's about to hand off
33 to the interpreter. That means your code must first compile correctly
34 for the debugger to work on it. Then when the interpreter starts up, it
35 preloads a Perl library file containing the debugger itself.
37 The program will halt I<right before> the first run-time executable
38 statement (but see below regarding compile-time statements) and ask you
39 to enter a debugger command. Contrary to popular expectations, whenever
40 the debugger halts and shows you a line of code, it always displays the
41 line it's I<about> to execute, rather than the one it has just executed.
43 Any command not recognized by the debugger is directly executed
44 (C<eval>'d) as Perl code in the current package. (The debugger uses the
45 DB package for its own state information.)
47 Leading white space before a command would cause the debugger to think
48 it's I<NOT> a debugger command but for Perl, so be careful not to do
51 =head2 Debugger Commands
53 The debugger understands the following commands:
59 Prints out a help message.
61 If you supply another debugger command as an argument to the C<h> command,
62 it prints out the description for just that command. The special
63 argument of C<h h> produces a more compact help listing, designed to fit
64 together on one screen.
66 If the output of the C<h> command (or any command, for that matter) scrolls
67 past your screen, either precede the command with a leading pipe symbol so
68 it's run through your pager, as in
72 You may change the pager which is used via C<O pager=...> command.
76 Same as C<print {$DB::OUT} expr> in the current package. In particular,
77 because this is just Perl's own B<print> function, this means that nested
78 data structures and objects are not dumped, unlike with the C<x> command.
80 The C<DB::OUT> filehandle is opened to F</dev/tty>, regardless of
81 where STDOUT may be redirected to.
85 Evaluates its expression in list context and dumps out the result
86 in a pretty-printed fashion. Nested data structures are printed out
87 recursively, unlike the C<print> function.
89 The details of printout are governed by multiple C<O>ptions.
93 Display all (or some) variables in package (defaulting to the C<main>
94 package) using a data pretty-printer (hashes show their keys and values so
95 you see what's what, control characters are made printable, etc.). Make
96 sure you don't put the type specifier (like C<$>) there, just the symbol
101 Use C<~pattern> and C<!pattern> for positive and negative regexps.
103 Nested data structures are printed out in a legible fashion, unlike
104 the C<print> function.
106 The details of printout are governed by multiple C<O>ptions.
110 Same as C<V currentpackage [vars]>.
114 Produce a stack backtrace. See below for details on its output.
118 Single step. Executes until it reaches the beginning of another
119 statement, descending into subroutine calls. If an expression is
120 supplied that includes function calls, it too will be single-stepped.
124 Next. Executes over subroutine calls, until it reaches the beginning
125 of the next statement. If an expression is supplied that includes
126 function calls, those functions will be executed with stops before
131 Repeat last C<n> or C<s> command.
135 Continue, optionally inserting a one-time-only breakpoint
136 at the specified line or subroutine.
140 List next window of lines.
144 List C<incr+1> lines starting at C<min>.
148 List lines C<min> through C<max>. C<l E<45>> is synonymous to C<E<45>>.
156 List first window of lines from subroutine. I<subname> may
157 be a variable which contains a code reference.
161 List previous window of lines.
165 List window (a few lines) around the current line.
169 Return debugger pointer to the last-executed line and
174 Switch to viewing a different file or eval statement. If C<filename>
175 is not a full filename as found in values of %INC, it is considered as
178 C<eval>ed strings (when accessible) are considered to be filenames:
179 C<f (eval 7)> and C<f eval 7\b> access the body of the 7th C<eval>ed string
180 (in the order of execution). The bodies of currently executed C<eval>
181 and of C<eval>ed strings which define subroutines are saved, thus are
182 accessible by this mechanism.
186 Search forwards for pattern; final / is optional.
190 Search backwards for pattern; final ? is optional.
194 List all breakpoints and actions.
198 List subroutine names [not] matching pattern.
202 Toggle trace mode (see also C<AutoTrace> C<O>ption).
206 Trace through execution of expr. For example:
209 Stack dump during die enabled outside of evals.
211 Loading DB routines from perl5db.pl patch level 0.94
212 Emacs support available.
214 Enter h or `h h' for help.
221 DB<3> t print foo() * bar()
222 main::((eval 172):3): print foo() + bar();
223 main::foo((eval 168):2):
224 main::bar((eval 170):2):
227 or, with the C<O>ption C<frame=2> set,
231 DB<5> t print foo() * bar()
241 =item b [line] [condition]
243 Set a breakpoint. If line is omitted, sets a breakpoint on the line
244 that is about to be executed. If a condition is specified, it's
245 evaluated each time the statement is reached and a breakpoint is taken
246 only if the condition is true. Breakpoints may be set on only lines
247 that begin an executable statement. Conditions don't use B<if>:
250 b 237 ++$count237 < 11
253 =item b subname [condition]
255 Set a breakpoint at the first line of the named subroutine. I<subname> may
256 be a variable which contains a code reference (in this case I<condition>
259 =item b postpone subname [condition]
261 Set breakpoint at first line of subroutine after it is compiled.
263 =item b load filename
265 Set breakpoint at the first executed line of the file. Filename should
266 be a full name as found in values of %INC.
268 =item b compile subname
270 Sets breakpoint at the first statement executed after the subroutine
275 Delete a breakpoint at the specified line. If line is omitted, deletes
276 the breakpoint on the line that is about to be executed.
280 Delete all installed breakpoints.
282 =item a [line] command
284 Set an action to be done before the line is executed.
285 The sequence of steps taken by the debugger is
287 1. check for a breakpoint at this line
288 2. print the line if necessary (tracing)
289 3. do any actions associated with that line
290 4. prompt user if at a breakpoint or in single-step
293 For example, this will print out $foo every time line
296 a 53 print "DB FOUND $foo\n"
300 Delete all installed actions.
304 Add a global watch-expression.
308 Delete all watch-expressions.
312 Continue until return from the current subroutine, and dump the return value.
314 =item O [opt[=val]] [opt"val"] [opt?]...
316 Set or query values of options. val defaults to 1. opt can
317 be abbreviated. Several options can be listed.
321 =item C<recallCommand>, C<ShellBang>
323 The characters used to recall command or spawn shell. By
324 default, these are both set to C<!>.
328 Program to use for output of pager-piped commands (those
329 beginning with a C<|> character.) By default,
330 C<$ENV{PAGER}> will be used.
334 Run Tk while prompting (with ReadLine).
336 =item C<signalLevel>, C<warnLevel>, C<dieLevel>
338 Level of verbosity. By default the debugger is in a sane verbose mode,
339 thus it will print backtraces on all the warnings and die-messages
340 which are going to be printed out, and will print a message when
341 interesting uncaught signals arrive.
343 To disable this behaviour, set these values to 0. If C<dieLevel> is 2,
344 then the messages which will be caught by surrounding C<eval> are also
349 Trace mode (similar to C<t> command, but can be put into
354 File or pipe to print line number info to. If it is a pipe (say,
355 C<|visual_perl_db>), then a short, "emacs like" message is used.
357 =item C<inhibit_exit>
359 If 0, allows I<stepping off> the end of the script.
363 affects printing of return value after C<r> command.
367 affects screen appearance of the command line (see L<Term::ReadLine>).
371 affects printing messages on entry and exit from subroutines. If
372 C<frame & 2> is false, messages are printed on entry only. (Printing
373 on exit may be useful if inter(di)spersed with other messages.)
375 If C<frame & 4>, arguments to functions are printed as well as the
376 context and caller info. If C<frame & 8>, overloaded C<stringify> and
377 C<tie>d C<FETCH> are enabled on the printed arguments. If C<frame &
378 16>, the return value from the subroutine is printed as well.
380 The length at which the argument list is truncated is governed by the
385 length at which the argument list is truncated when C<frame> option's
390 The following options affect what happens with C<V>, C<X>, and C<x>
395 =item C<arrayDepth>, C<hashDepth>
397 Print only first N elements ('' for all).
399 =item C<compactDump>, C<veryCompact>
401 Change style of array and hash dump. If C<compactDump>, short array
402 may be printed on one line.
406 Whether to print contents of globs.
410 Dump arrays holding debugged files.
412 =item C<DumpPackages>
414 Dump symbol tables of packages.
418 Dump contents of "reused" addresses.
420 =item C<quote>, C<HighBit>, C<undefPrint>
422 Change style of string dump. Default value of C<quote> is C<auto>, one
423 can enable either double-quotish dump, or single-quotish by setting it
424 to C<"> or C<'>. By default, characters with high bit set are printed
429 I<very> rudimentally per-package memory usage dump. Calculates total
430 size of strings in variables in the package.
434 During startup options are initialized from C<$ENV{PERLDB_OPTS}>.
435 You can put additional initialization options C<TTY>, C<noTTY>,
436 C<ReadLine>, and C<NonStop> there.
440 &parse_options("NonStop=1 LineInfo=db.out AutoTrace");
442 The script will run without human intervention, putting trace information
443 into the file I<db.out>. (If you interrupt it, you would better reset
444 C<LineInfo> to something "interactive"!)
450 The TTY to use for debugging I/O.
454 If set, goes in C<NonStop> mode, and would not connect to a TTY. If
455 interrupt (or if control goes to debugger via explicit setting of
456 $DB::signal or $DB::single from the Perl script), connects to a TTY
457 specified by the C<TTY> option at startup, or to a TTY found at
458 runtime using C<Term::Rendezvous> module of your choice.
460 This module should implement a method C<new> which returns an object
461 with two methods: C<IN> and C<OUT>, returning two filehandles to use
462 for debugging input and output correspondingly. Method C<new> may
463 inspect an argument which is a value of C<$ENV{PERLDB_NOTTY}> at
464 startup, or is C<"/tmp/perldbtty$$"> otherwise.
468 If false, readline support in debugger is disabled, so you can debug
469 ReadLine applications.
473 If set, debugger goes into noninteractive mode until interrupted, or
474 programmatically by setting $DB::signal or $DB::single.
478 Here's an example of using the C<$ENV{PERLDB_OPTS}> variable:
480 $ PERLDB_OPTS="N f=2" perl -d myprogram
482 will run the script C<myprogram> without human intervention, printing
483 out the call tree with entry and exit points. Note that C<N f=2> is
484 equivalent to C<NonStop=1 frame=2>. Note also that at the moment when
485 this documentation was written all the options to the debugger could
486 be uniquely abbreviated by the first letter (with exception of
489 Other examples may include
491 $ PERLDB_OPTS="N f A L=listing" perl -d myprogram
493 - runs script noninteractively, printing info on each entry into a
494 subroutine and each executed line into the file F<listing>. (If you
495 interrupt it, you would better reset C<LineInfo> to something
499 $ env "PERLDB_OPTS=R=0 TTY=/dev/ttyc" perl -d myprogram
501 may be useful for debugging a program which uses C<Term::ReadLine>
502 itself. Do not forget detach shell from the TTY in the window which
503 corresponds to F</dev/ttyc>, say, by issuing a command like
507 See L<"Debugger Internals"> below for more details.
509 =item E<lt> [ command ]
511 Set an action (Perl command) to happen before every debugger prompt.
512 A multi-line command may be entered by backslashing the newlines. If
513 C<command> is missing, resets the list of actions.
515 =item E<lt>E<lt> command
517 Add an action (Perl command) to happen before every debugger prompt.
518 A multi-line command may be entered by backslashing the newlines.
522 Set an action (Perl command) to happen after the prompt when you've
523 just given a command to return to executing the script. A multi-line
524 command may be entered by backslashing the newlines. If C<command> is
525 missing, resets the list of actions.
527 =item E<gt>E<gt> command
529 Adds an action (Perl command) to happen after the prompt when you've
530 just given a command to return to executing the script. A multi-line
531 command may be entered by backslashing the newlines.
535 Set an action (debugger command) to happen before every debugger prompt.
536 A multi-line command may be entered by backslashing the newlines. If
537 C<command> is missing, resets the list of actions.
541 Add an action (debugger command) to happen before every debugger prompt.
542 A multi-line command may be entered by backslashing the newlines.
546 Redo a previous command (default previous command).
550 Redo number'th-to-last command.
554 Redo last command that started with pattern.
555 See C<O recallCommand>, too.
559 Run cmd in a subprocess (reads from DB::IN, writes to DB::OUT)
560 See C<O shellBang> too.
564 Display last n commands. Only commands longer than one character are
565 listed. If number is omitted, lists them all.
569 Quit. ("quit" doesn't work for this.) This is the only supported way
570 to exit the debugger, though typing C<exit> twice may do it too.
572 Set an C<O>ption C<inhibit_exit> to 0 if you want to be able to I<step
573 off> the end the script. You may also need to set $finished to 0 at
574 some moment if you want to step through global destruction.
578 Restart the debugger by B<exec>ing a new session. It tries to maintain
579 your history across this, but internal settings and command line options
582 Currently the following setting are preserved: history, breakpoints,
583 actions, debugger C<O>ptions, and the following command line
584 options: B<-w>, B<-I>, and B<-e>.
588 Run debugger command, piping DB::OUT to current pager.
592 Same as C<|dbcmd> but DB::OUT is temporarily B<select>ed as well.
593 Often used with commands that would otherwise produce long
598 =item = [alias value]
600 Define a command alias, like
604 or list current aliases.
608 Execute command as a Perl statement. A missing semicolon will be
613 The expression is evaluated, and the methods which may be applied to
614 the result are listed.
618 The methods which may be applied to objects in the C<package> are listed.
622 =head2 Debugger input/output
628 The debugger prompt is something like
636 where that number is the command number, which you'd use to access with
637 the builtin B<csh>-like history mechanism, e.g., C<!17> would repeat
638 command number 17. The number of angle brackets indicates the depth of
639 the debugger. You could get more than one set of brackets, for example, if
640 you'd already at a breakpoint and then printed out the result of a
641 function call that itself also has a breakpoint, or you step into an
642 expression via C<s/n/t expression> command.
644 =item Multiline commands
646 If you want to enter a multi-line command, such as a subroutine
647 definition with several statements, or a format, you may escape the
648 newline that would normally end the debugger command with a backslash.
652 cont: print "ok\n"; \
659 Note that this business of escaping a newline is specific to interactive
660 commands typed into the debugger.
662 =item Stack backtrace
664 Here's an example of what a stack backtrace via C<T> command might
667 $ = main::infested called from file `Ambulation.pm' line 10
668 @ = Ambulation::legs(1, 2, 3, 4) called from file `camel_flea' line 7
669 $ = main::pests('bactrian', 4) called from file `camel_flea' line 4
671 The left-hand character up there tells whether the function was called
672 in a scalar or list context (we bet you can tell which is which). What
673 that says is that you were in the function C<main::infested> when you ran
674 the stack dump, and that it was called in a scalar context from line 10
675 of the file I<Ambulation.pm>, but without any arguments at all, meaning
676 it was called as C<&infested>. The next stack frame shows that the
677 function C<Ambulation::legs> was called in a list context from the
678 I<camel_flea> file with four arguments. The last stack frame shows that
679 C<main::pests> was called in a scalar context, also from I<camel_flea>,
682 Note that if you execute C<T> command from inside an active C<use>
683 statement, the backtrace will contain both C<require>
684 frame and an C<eval>) frame.
688 Listing given via different flavors of C<l> command looks like this:
692 102:b @isa{@i,$pack} = ()
693 103 if(exists $i{$prevpack} || exists $isa{$pack});
697 107==> if(exists $isa{$pack});
699 109:a if ($extra-- > 0) {
700 110: %isa = ($pack,1);
702 Note that the breakable lines are marked with C<:>, lines with
703 breakpoints are marked by C<b>, with actions by C<a>, and the
704 next executed line is marked by C<==E<gt>>.
708 When C<frame> option is set, debugger would print entered (and
709 optionally exited) subroutines in different styles.
711 What follows is the start of the listing of
713 env "PERLDB_OPTS=f=n N" perl -d -V
715 for different values of C<n>:
722 entering Config::BEGIN
723 Package lib/Exporter.pm.
725 Package lib/Config.pm.
726 entering Config::TIEHASH
727 entering Exporter::import
728 entering Exporter::export
729 entering Config::myconfig
730 entering Config::FETCH
731 entering Config::FETCH
732 entering Config::FETCH
733 entering Config::FETCH
738 entering Config::BEGIN
739 Package lib/Exporter.pm.
742 Package lib/Config.pm.
743 entering Config::TIEHASH
744 exited Config::TIEHASH
745 entering Exporter::import
746 entering Exporter::export
747 exited Exporter::export
748 exited Exporter::import
750 entering Config::myconfig
751 entering Config::FETCH
753 entering Config::FETCH
755 entering Config::FETCH
759 in $=main::BEGIN() from /dev/nul:0
760 in $=Config::BEGIN() from lib/Config.pm:2
761 Package lib/Exporter.pm.
763 Package lib/Config.pm.
764 in $=Config::TIEHASH('Config') from lib/Config.pm:644
765 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/nul:0
766 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from li
767 in @=Config::myconfig() from /dev/nul:0
768 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
769 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
770 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
771 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
772 in $=Config::FETCH(ref(Config), 'osname') from lib/Config.pm:574
773 in $=Config::FETCH(ref(Config), 'osvers') from lib/Config.pm:574
777 in $=main::BEGIN() from /dev/nul:0
778 in $=Config::BEGIN() from lib/Config.pm:2
779 Package lib/Exporter.pm.
781 out $=Config::BEGIN() from lib/Config.pm:0
782 Package lib/Config.pm.
783 in $=Config::TIEHASH('Config') from lib/Config.pm:644
784 out $=Config::TIEHASH('Config') from lib/Config.pm:644
785 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/nul:0
786 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
787 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/
788 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/nul:0
789 out $=main::BEGIN() from /dev/nul:0
790 in @=Config::myconfig() from /dev/nul:0
791 in $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
792 out $=Config::FETCH(ref(Config), 'package') from lib/Config.pm:574
793 in $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
794 out $=Config::FETCH(ref(Config), 'baserev') from lib/Config.pm:574
795 in $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
796 out $=Config::FETCH(ref(Config), 'PERL_VERSION') from lib/Config.pm:574
797 in $=Config::FETCH(ref(Config), 'PERL_SUBVERSION') from lib/Config.pm:574
801 in $=main::BEGIN() from /dev/nul:0
802 in $=Config::BEGIN() from lib/Config.pm:2
803 Package lib/Exporter.pm.
805 out $=Config::BEGIN() from lib/Config.pm:0
806 Package lib/Config.pm.
807 in $=Config::TIEHASH('Config') from lib/Config.pm:644
808 out $=Config::TIEHASH('Config') from lib/Config.pm:644
809 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/nul:0
810 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
811 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/E
812 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/nul:0
813 out $=main::BEGIN() from /dev/nul:0
814 in @=Config::myconfig() from /dev/nul:0
815 in $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
816 out $=Config::FETCH('Config=HASH(0x1aa444)', 'package') from lib/Config.pm:574
817 in $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
818 out $=Config::FETCH('Config=HASH(0x1aa444)', 'baserev') from lib/Config.pm:574
822 in $=CODE(0x15eca4)() from /dev/null:0
823 in $=CODE(0x182528)() from lib/Config.pm:2
824 Package lib/Exporter.pm.
825 out $=CODE(0x182528)() from lib/Config.pm:0
826 scalar context return from CODE(0x182528): undef
827 Package lib/Config.pm.
828 in $=Config::TIEHASH('Config') from lib/Config.pm:628
829 out $=Config::TIEHASH('Config') from lib/Config.pm:628
830 scalar context return from Config::TIEHASH: empty hash
831 in $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
832 in $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
833 out $=Exporter::export('Config', 'main', 'myconfig', 'config_vars') from lib/Exporter.pm:171
834 scalar context return from Exporter::export: ''
835 out $=Exporter::import('Config', 'myconfig', 'config_vars') from /dev/null:0
836 scalar context return from Exporter::import: ''
841 In all the cases indentation of lines shows the call tree, if bit 2 of
842 C<frame> is set, then a line is printed on exit from a subroutine as
843 well, if bit 4 is set, then the arguments are printed as well as the
844 caller info, if bit 8 is set, the arguments are printed even if they
845 are tied or references, if bit 16 is set, the return value is printed
848 When a package is compiled, a line like this
852 is printed with proper indentation.
856 =head2 Debugging compile-time statements
858 If you have any compile-time executable statements (code within a BEGIN
859 block or a C<use> statement), these will C<NOT> be stopped by debugger,
860 although C<require>s will (and compile-time statements can be traced
861 with C<AutoTrace> option set in C<PERLDB_OPTS>). From your own Perl
862 code, however, you can
863 transfer control back to the debugger using the following statement,
864 which is harmless if the debugger is not running:
868 If you set C<$DB::single> to the value 2, it's equivalent to having
869 just typed the C<n> command, whereas a value of 1 means the C<s>
870 command. The C<$DB::trace> variable should be set to 1 to simulate
871 having typed the C<t> command.
873 Another way to debug compile-time code is to start debugger, set a
874 breakpoint on I<load> of some module thusly
876 DB<7> b load f:/perllib/lib/Carp.pm
877 Will stop on load of `f:/perllib/lib/Carp.pm'.
879 and restart debugger by C<R> command (if possible). One can use C<b
880 compile subname> for the same purpose.
882 =head2 Debugger Customization
884 Most probably you do not want to modify the debugger, it contains enough
885 hooks to satisfy most needs. You may change the behaviour of debugger
886 from the debugger itself, using C<O>ptions, from the command line via
887 C<PERLDB_OPTS> environment variable, and from I<customization files>.
889 You can do some customization by setting up a F<.perldb> file which
890 contains initialization code. For instance, you could make aliases
891 like these (the last one is one people expect to be there):
893 $DB::alias{'len'} = 's/^len(.*)/p length($1)/';
894 $DB::alias{'stop'} = 's/^stop (at|in)/b/';
895 $DB::alias{'ps'} = 's/^ps\b/p scalar /';
896 $DB::alias{'quit'} = 's/^quit(\s*)/exit\$/';
898 One changes options from F<.perldb> file via calls like this one;
900 parse_options("NonStop=1 LineInfo=db.out AutoTrace=1 frame=2");
902 (the code is executed in the package C<DB>). Note that F<.perldb> is
903 processed before processing C<PERLDB_OPTS>. If F<.perldb> defines the
904 subroutine C<afterinit>, it is called after all the debugger
905 initialization ends. F<.perldb> may be contained in the current
906 directory, or in the C<LOGDIR>/C<HOME> directory.
908 If you want to modify the debugger, copy F<perl5db.pl> from the Perl
909 library to another name and modify it as necessary. You'll also want
910 to set your C<PERL5DB> environment variable to say something like this:
912 BEGIN { require "myperl5db.pl" }
914 As the last resort, one can use C<PERL5DB> to customize debugger by
915 directly setting internal variables or calling debugger functions.
917 =head2 Readline Support
919 As shipped, the only command line history supplied is a simplistic one
920 that checks for leading exclamation points. However, if you install
921 the Term::ReadKey and Term::ReadLine modules from CPAN, you will
922 have full editing capabilities much like GNU I<readline>(3) provides.
923 Look for these in the F<modules/by-module/Term> directory on CPAN.
925 A rudimentary command line completion is also available.
926 Unfortunately, the names of lexical variables are not available for
929 =head2 Editor Support for Debugging
931 If you have GNU B<emacs> installed on your system, it can interact with
932 the Perl debugger to provide an integrated software development
933 environment reminiscent of its interactions with C debuggers.
935 Perl is also delivered with a start file for making B<emacs> act like a
936 syntax-directed editor that understands (some of) Perl's syntax. Look in
937 the I<emacs> directory of the Perl source distribution.
939 (Historically, a similar setup for interacting with B<vi> and the
940 X11 window system had also been available, but at the time of this
941 writing, no debugger support for B<vi> currently exists.)
943 =head2 The Perl Profiler
945 If you wish to supply an alternative debugger for Perl to run, just
946 invoke your script with a colon and a package argument given to the B<-d>
947 flag. One of the most popular alternative debuggers for Perl is
948 B<DProf>, the Perl profiler. As of this writing, B<DProf> is not
949 included with the standard Perl distribution, but it is expected to
950 be included soon, for certain values of "soon".
952 Meanwhile, you can fetch the Devel::Dprof module from CPAN. Assuming
953 it's properly installed on your system, to profile your Perl program in
954 the file F<mycode.pl>, just type:
956 perl -d:DProf mycode.pl
958 When the script terminates the profiler will dump the profile information
959 to a file called F<tmon.out>. A tool like B<dprofpp> (also supplied with
960 the Devel::DProf package) can be used to interpret the information which is
963 =head2 Debugger support in perl
965 When you call the B<caller> function (see L<perlfunc/caller>) from the
966 package DB, Perl sets the array @DB::args to contain the arguments the
967 corresponding stack frame was called with.
969 If perl is run with B<-d> option, the following additional features
970 are enabled (cf. L<perlvar/$^P>):
976 Perl inserts the contents of C<$ENV{PERL5DB}> (or C<BEGIN {require
977 'perl5db.pl'}> if not present) before the first line of the
982 The array C<@{"_E<lt>$filename"}> is the line-by-line contents of
983 $filename for all the compiled files. Same for C<eval>ed strings which
984 contain subroutines, or which are currently executed. The $filename
985 for C<eval>ed strings looks like C<(eval 34)>.
989 The hash C<%{"_E<lt>$filename"}> contains breakpoints and action (it is
990 keyed by line number), and individual entries are settable (as opposed
991 to the whole hash). Only true/false is important to Perl, though the
992 values used by F<perl5db.pl> have the form
993 C<"$break_condition\0$action">. Values are magical in numeric context:
994 they are zeros if the line is not breakable.
996 Same for evaluated strings which contain subroutines, or which are
997 currently executed. The $filename for C<eval>ed strings looks like
1002 The scalar C<${"_E<lt>$filename"}> contains C<"_E<lt>$filename">. Same for
1003 evaluated strings which contain subroutines, or which are currently
1004 executed. The $filename for C<eval>ed strings looks like C<(eval
1009 After each C<require>d file is compiled, but before it is executed,
1010 C<DB::postponed(*{"_E<lt>$filename"})> is called (if subroutine
1011 C<DB::postponed> exists). Here the $filename is the expanded name of
1012 the C<require>d file (as found in values of %INC).
1016 After each subroutine C<subname> is compiled existence of
1017 C<$DB::postponed{subname}> is checked. If this key exists,
1018 C<DB::postponed(subname)> is called (if subroutine C<DB::postponed>
1023 A hash C<%DB::sub> is maintained, with keys being subroutine names,
1024 values having the form C<filename:startline-endline>. C<filename> has
1025 the form C<(eval 31)> for subroutines defined inside C<eval>s.
1029 When execution of the application reaches a place that can have
1030 a breakpoint, a call to C<DB::DB()> is performed if any one of
1031 variables $DB::trace, $DB::single, or $DB::signal is true. (Note that
1032 these variables are not C<local>izable.) This feature is disabled when
1033 the control is inside C<DB::DB()> or functions called from it (unless
1034 C<$^D & (1E<lt>E<lt>30)>).
1038 When execution of the application reaches a subroutine call, a call
1039 to C<&DB::sub>(I<args>) is performed instead, with C<$DB::sub> being
1040 the name of the called subroutine. (Unless the subroutine is compiled
1041 in the package C<DB>.)
1045 Note that if C<&DB::sub> needs some external data to be setup for it
1046 to work, no subroutine call is possible until this is done. For the
1047 standard debugger C<$DB::deep> (how many levels of recursion deep into
1048 the debugger you can go before a mandatory break) gives an example of
1051 The minimal working debugger consists of one line
1055 which is quite handy as contents of C<PERL5DB> environment
1058 env "PERL5DB=sub DB::DB {}" perl -d your-script
1060 Another (a little bit more useful) minimal debugger can be created
1061 with the only line being
1063 sub DB::DB {print ++$i; scalar <STDIN>}
1065 This debugger would print the sequential number of encountered
1066 statement, and would wait for your C<CR> to continue.
1068 The following debugger is quite functional:
1073 sub sub {print ++$i, " $sub\n"; &$sub}
1076 It prints the sequential number of subroutine call and the name of the
1077 called subroutine. Note that C<&DB::sub> should be compiled into the
1080 =head2 Debugger Internals
1082 At the start, the debugger reads your rc file (F<./.perldb> or
1083 F<~/.perldb> under Unix), which can set important options. This file may
1084 define a subroutine C<&afterinit> to be executed after the debugger is
1087 After the rc file is read, the debugger reads environment variable
1088 PERLDB_OPTS and parses it as a rest of C<O ...> line in debugger prompt.
1090 It also maintains magical internal variables, such as C<@DB::dbline>,
1091 C<%DB::dbline>, which are aliases for C<@{"::_<current_file"}>
1092 C<%{"::_<current_file"}>. Here C<current_file> is the currently
1093 selected (with the debugger's C<f> command, or by flow of execution)
1096 Some functions are provided to simplify customization. See L<"Debugger
1097 Customization"> for description of C<DB::parse_options(string)>. The
1098 function C<DB::dump_trace(skip[, count])> skips the specified number
1099 of frames, and returns a list containing info about the caller
1100 frames (all if C<count> is missing). Each entry is a hash with keys
1101 C<context> (C<$> or C<@>), C<sub> (subroutine name, or info about
1102 eval), C<args> (C<undef> or a reference to an array), C<file>, and
1105 The function C<DB::print_trace(FH, skip[, count[, short]])> prints
1106 formatted info about caller frames. The last two functions may be
1107 convenient as arguments to C<E<lt>>, C<E<lt>E<lt>> commands.
1109 =head2 Other resources
1111 You did try the B<-w> switch, didn't you?
1115 You cannot get the stack frame information or otherwise debug functions
1116 that were not compiled by Perl, such as C or C++ extensions.
1118 If you alter your @_ arguments in a subroutine (such as with B<shift>
1119 or B<pop>, the stack backtrace will not show the original values.
1121 =head1 Debugging Perl memory usage
1123 Perl is I<very> frivolous with memory. There is a saying that to
1124 estimate memory usage of Perl, assume a reasonable algorithm of
1125 allocation, and multiply your estimates by 10. This is not absolutely
1126 true, but may give you a good grasp of what happens.
1128 Say, an integer cannot take less than 20 bytes of memory, a float
1129 cannot take less than 24 bytes, a string cannot take less than 32
1130 bytes (all these examples assume 32-bit architectures, the result are
1131 much worse on 64-bit architectures). If a variable is accessed in two
1132 of three different ways (which require an integer, a float, or a
1133 string), the memory footprint may increase by another 20 bytes. A
1134 sloppy malloc() implementation will make these numbers yet more.
1136 On the opposite end of the scale, a declaration like
1140 may take (on some versions of perl) up to 500 bytes of memory.
1142 Off-the-cuff anecdotal estimates of a code bloat give a factor around
1143 8. This means that the compiled form of reasonable (commented
1144 indented etc.) code will take approximately 8 times more than the
1145 disk space the code takes.
1147 There are two Perl-specific ways to analyze the memory usage:
1148 $ENV{PERL_DEBUG_MSTATS} and B<-DL> switch. First one is available
1149 only if perl is compiled with Perl's malloc(), the second one only if
1150 Perl compiled with C<-DDEBUGGING> (as with giving C<-D optimise=-g>
1151 option to F<Configure>).
1153 =head2 Using C<$ENV{PERL_DEBUG_MSTATS}>
1155 If your perl is using Perl's malloc(), and compiled with correct
1156 switches (this is the default), then it will print memory usage
1157 statistics after compiling your code (if C<$ENV{PERL_DEBUG_MSTATS}> >
1158 1), and before termination of the script (if
1159 C<$ENV{PERL_DEBUG_MSTATS}> >= 1). The report format is similar to one
1160 in the following example:
1162 env PERL_DEBUG_MSTATS=2 perl -e "require Carp"
1163 Memory allocation statistics after compilation: (buckets 4(4)..8188(8192)
1164 14216 free: 130 117 28 7 9 0 2 2 1 0 0
1166 60924 used: 125 137 161 55 7 8 6 16 2 0 1
1168 Total sbrk(): 77824/21:119. Odd ends: pad+heads+chain+tail: 0+636+0+2048.
1169 Memory allocation statistics after execution: (buckets 4(4)..8188(8192)
1170 30888 free: 245 78 85 13 6 2 1 3 2 0 1
1172 175816 used: 265 176 1112 111 26 22 11 27 2 1 1
1174 Total sbrk(): 215040/47:145. Odd ends: pad+heads+chain+tail: 0+2192+0+6144.
1176 It is possible to ask for such a statistic at arbitrary moment by
1177 using Devel::Peek::mstats() (module Devel::Peek is available on CPAN).
1179 Here is the explanation of different parts of the format:
1183 =item C<buckets SMALLEST(APPROX)..GREATEST(APPROX)>
1185 Perl's malloc() uses bucketed allocations. Every request is rounded
1186 up to the closest bucket size available, and a bucket of these size is
1187 taken from the pool of the buckets of this size.
1189 The above line describes limits of buckets currently in use. Each
1190 bucket has two sizes: memory footprint, and the maximal size of user
1191 data which may be put into this bucket. Say, in the above example the
1192 smallest bucket is both sizes 4. The biggest bucket has usable size
1193 8188, and the memory footprint 8192.
1195 With debugging Perl some buckets may have negative usable size. This
1196 means that these buckets cannot (and will not) be used. For greater
1197 buckets the memory footprint may be one page greater than a power of
1198 2. In such a case the corresponding power of two is printed instead
1199 in the C<APPROX> field above.
1203 The following 1 or 2 rows of numbers correspond to the number of
1204 buckets of each size between C<SMALLEST> and C<GREATEST>. In the
1205 first row the sizes (memory footprints) of buckets are powers of two
1206 (or possibly one page greater). In the second row (if present) the
1207 memory footprints of the buckets are between memory footprints of two
1210 Say, with the above example the memory footprints are (with current
1213 free: 8 16 32 64 128 256 512 1024 2048 4096 8192
1216 With non-C<DEBUGGING> perl the buckets starting from C<128>-long ones
1217 have 4-byte overhead, thus 8192-long bucket may take up to
1218 8188-byte-long allocations.
1220 =item C<Total sbrk(): SBRKed/SBRKs:CONTINUOUS>
1222 The first two fields give the total amount of memory perl sbrk()ed,
1223 and number of sbrk()s used. The third number is what perl thinks
1224 about continuity of returned chunks. As far as this number is
1225 positive, malloc() will assume that it is probable that sbrk() will
1226 provide continuous memory.
1228 The amounts sbrk()ed by external libraries is not counted.
1232 The amount of sbrk()ed memory needed to keep buckets aligned.
1234 =item C<heads: 2192>
1236 While memory overhead of bigger buckets is kept inside the bucket, for
1237 smaller buckets it is kept in separate areas. This field gives the
1238 total size of these areas.
1242 malloc() may want to subdivide a bigger bucket into smaller buckets.
1243 If only a part of the deceased-bucket is left non-subdivided, the rest
1244 is kept as an element of a linked list. This field gives the total
1245 size of these chunks.
1249 To minimize amount of sbrk()s malloc() asks for more memory. This
1250 field gives the size of the yet-unused part, which is sbrk()ed, but
1255 =head2 Example of using B<-DL> switch
1257 Below we show how to analyse memory usage by
1259 do 'lib/auto/POSIX/autosplit.ix';
1261 The file in question contains a header and 146 lines similar to
1265 B<Note:> I<the discussion below supposes 32-bit architecture. In the
1266 newer versions of perl the memory usage of the constructs discussed
1267 here is much improved, but the story discussed below is a real-life
1268 story. This story is very terse, and assumes more than cursory
1269 knowledge of Perl internals.>
1271 Here is the itemized list of Perl allocations performed during parsing
1274 !!! "after" at test.pl line 3.
1275 Id subtot 4 8 12 16 20 24 28 32 36 40 48 56 64 72 80 80+
1276 0 02 13752 . . . . 294 . . . . . . . . . . 4
1277 0 54 5545 . . 8 124 16 . . . 1 1 . . . . . 3
1278 5 05 32 . . . . . . . 1 . . . . . . . .
1279 6 02 7152 . . . . . . . . . . 149 . . . . .
1280 7 02 3600 . . . . . 150 . . . . . . . . . .
1281 7 03 64 . -1 . 1 . . 2 . . . . . . . . .
1282 7 04 7056 . . . . . . . . . . . . . . . 7
1283 7 17 38404 . . . . . . . 1 . . 442 149 . . 147 .
1284 9 03 2078 17 249 32 . . . . 2 . . . . . . . .
1287 To see this list insert two C<warn('!...')> statements around the call:
1290 do 'lib/auto/POSIX/autosplit.ix';
1291 warn('!!! "after"');
1293 and run it with B<-DL> option. The first warn() will print memory
1294 allocation info before the parsing of the file, and will memorize the
1295 statistics at this point (we ignore what it prints). The second warn()
1296 will print increments w.r.t. this memorized statistics. This is the
1299 Different I<Id>s on the left correspond to different subsystems of
1300 perl interpreter, they are just first argument given to perl memory
1301 allocation API New(). To find what C<9 03> means C<grep> the perl
1302 source for C<903>. You will see that it is F<util.c>, function
1303 savepvn(). This function is used to store a copy of existing chunk of
1304 memory. Using C debugger, one can see that it is called either
1305 directly from gv_init(), or via sv_magic(), and gv_init() is called
1306 from gv_fetchpv() - which is called from newSUB().
1308 B<Note:> to reach this place in debugger and skip all the calls to
1309 savepvn during the compilation of the main script, set a C breakpoint
1310 in Perl_warn(), C<continue> this point is reached, I<then> set
1311 breakpoint in Perl_savepvn(). Note that you may need to skip a
1312 handful of Perl_savepvn() which do not correspond to mass production
1313 of CVs (there are more C<903> allocations than 146 similar lines of
1314 F<lib/auto/POSIX/autosplit.ix>). Note also that C<Perl_> prefixes are
1315 added by macroization code in perl header files to avoid conflicts
1316 with external libraries.
1318 Anyway, we see that C<903> ids correspond to creation of globs, twice
1319 per glob - for glob name, and glob stringification magic.
1321 Here are explanations for other I<Id>s above:
1327 is for creation of bigger C<XPV*> structures. In the above case it
1328 creates 3 C<AV> per subroutine, one for a list of lexical variable
1329 names, one for a scratchpad (which contains lexical variables and
1330 C<targets>), and one for the array of scratchpads needed for
1333 It also creates a C<GV> and a C<CV> per subroutine (all called from
1338 Creates C array corresponding to the C<AV> of scratchpads, and the
1339 scratchpad itself (the first fake entry of this scratchpad is created
1340 though the subroutine itself is not defined yet).
1342 It also creates C arrays to keep data for the stash (this is one HV,
1343 but it grows, thus there are 4 big allocations: the big chunks are not
1344 freed, but are kept as additional arenas for C<SV> allocations).
1348 creates a C<HEK> for the name of the glob for the subroutine (this
1349 name is a key in a I<stash>).
1351 Big allocations with this I<Id> correspond to allocations of new
1352 arenas to keep C<HE>.
1356 creates a C<GP> for the glob for the subroutine.
1360 creates the C<MAGIC> for the glob for the subroutine.
1364 creates I<arenas> which keep SVs.
1368 =head2 B<-DL> details
1370 If Perl is run with B<-DL> option, then warn()s which start with `!'
1371 behave specially. They print a list of I<categories> of memory
1372 allocations, and statistics of allocations of different sizes for
1375 If warn() string starts with
1381 print changed categories only, print the differences in counts of allocations;
1385 print grown categories only; print the absolute values of counts, and totals;
1389 print nonempty categories, print the absolute values of counts and totals.
1393 =head2 Limitations of B<-DL> statistic
1395 If an extension or an external library does not use Perl API to
1396 allocate memory, these allocations are not counted.
1398 =head1 Debugging regular expressions
1400 There are two ways to enable debugging output for regular expressions.
1402 If your perl is compiled with C<-DDEBUGGING>, you may use the
1403 B<-Dr> flag on the command line.
1405 Otherwise, one can C<use re 'debug'>, which has effects both at
1406 compile time, and at run time (and is I<not> lexically scoped).
1408 =head2 Compile-time output
1410 The debugging output for the compile time looks like this:
1412 compiling RE `[bc]d(ef*g)+h[ij]k$'
1416 13: CURLYX {1,32767}(27)
1430 anchored `de' at 1 floating `gh' at 3..2147483647 (checking floating)
1431 stclass `ANYOF' minlen 7
1433 The first line shows the pre-compiled form of the regexp, and the
1434 second shows the size of the compiled form (in arbitrary units,
1435 usually 4-byte words) and the label I<id> of the first node which
1438 The last line (split into two lines in the above) contains the optimizer
1439 info. In the example shown, the optimizer found that the match
1440 should contain a substring C<de> at the offset 1, and substring C<gh>
1441 at some offset between 3 and infinity. Moreover, when checking for
1442 these substrings (to abandon impossible matches quickly) it will check
1443 for the substring C<gh> before checking for the substring C<de>. The
1444 optimizer may also use the knowledge that the match starts (at the
1445 C<first> I<id>) with a character class, and the match cannot be
1446 shorter than 7 chars.
1448 The fields of interest which may appear in the last line are
1452 =item C<anchored> I<STRING> C<at> I<POS>
1454 =item C<floating> I<STRING> C<at> I<POS1..POS2>
1458 =item C<matching floating/anchored>
1460 which substring to check first;
1464 the minimal length of the match;
1466 =item C<stclass> I<TYPE>
1468 The type of the first matching node.
1472 which advises to not scan for the found substrings;
1476 which says that the optimizer info is in fact all that the regular
1477 expression contains (thus one does not need to enter the RE engine at
1482 if the pattern contains C<\G>;
1486 if the pattern starts with a repeated char (as in C<x+y>);
1490 if the pattern starts with C<.*>;
1494 if the pattern contain eval-groups (see L<perlre/(?{ code })>);
1496 =item C<anchored(TYPE)>
1499 match only at a handful of places (with C<TYPE> being
1500 C<BOL>, C<MBOL>, or C<GPOS>, see the table below).
1504 If a substring is known to match at end-of-line only, it may be
1505 followed by C<$>, as in C<floating `k'$>.
1507 The optimizer-specific info is used to avoid entering (a slow) RE
1508 engine on strings which will definitely not match. If C<isall> flag
1509 is set, a call to the RE engine may be avoided even when optimizer
1510 found an appropriate place for the match.
1512 The rest of the output contains the list of I<nodes> of the compiled
1513 form of the RE. Each line has format
1515 C< >I<id>: I<TYPE> I<OPTIONAL-INFO> (I<next-id>)
1517 =head2 Types of nodes
1519 Here is the list of possible types with short descriptions:
1521 # TYPE arg-description [num-args] [longjump-len] DESCRIPTION
1524 END no End of program.
1525 SUCCEED no Return from a subroutine, basically.
1528 BOL no Match "" at beginning of line.
1529 MBOL no Same, assuming multiline.
1530 SBOL no Same, assuming singleline.
1531 EOS no Match "" at end of string.
1532 EOL no Match "" at end of line.
1533 MEOL no Same, assuming multiline.
1534 SEOL no Same, assuming singleline.
1535 BOUND no Match "" at any word boundary
1536 BOUNDL no Match "" at any word boundary
1537 NBOUND no Match "" at any word non-boundary
1538 NBOUNDL no Match "" at any word non-boundary
1539 GPOS no Matches where last m//g left off.
1541 # [Special] alternatives
1542 ANY no Match any one character (except newline).
1543 SANY no Match any one character.
1544 ANYOF sv Match character in (or not in) this class.
1545 ALNUM no Match any alphanumeric character
1546 ALNUML no Match any alphanumeric char in locale
1547 NALNUM no Match any non-alphanumeric character
1548 NALNUML no Match any non-alphanumeric char in locale
1549 SPACE no Match any whitespace character
1550 SPACEL no Match any whitespace char in locale
1551 NSPACE no Match any non-whitespace character
1552 NSPACEL no Match any non-whitespace char in locale
1553 DIGIT no Match any numeric character
1554 NDIGIT no Match any non-numeric character
1556 # BRANCH The set of branches constituting a single choice are hooked
1557 # together with their "next" pointers, since precedence prevents
1558 # anything being concatenated to any individual branch. The
1559 # "next" pointer of the last BRANCH in a choice points to the
1560 # thing following the whole choice. This is also where the
1561 # final "next" pointer of each individual branch points; each
1562 # branch starts with the operand node of a BRANCH node.
1564 BRANCH node Match this alternative, or the next...
1566 # BACK Normal "next" pointers all implicitly point forward; BACK
1567 # exists to make loop structures possible.
1569 BACK no Match "", "next" ptr points backward.
1572 EXACT sv Match this string (preceded by length).
1573 EXACTF sv Match this string, folded (prec. by length).
1574 EXACTFL sv Match this string, folded in locale (w/len).
1577 NOTHING no Match empty string.
1578 # A variant of above which delimits a group, thus stops optimizations
1579 TAIL no Match empty string. Can jump here from outside.
1581 # STAR,PLUS '?', and complex '*' and '+', are implemented as circular
1582 # BRANCH structures using BACK. Simple cases (one character
1583 # per match) are implemented with STAR and PLUS for speed
1584 # and to minimize recursive plunges.
1586 STAR node Match this (simple) thing 0 or more times.
1587 PLUS node Match this (simple) thing 1 or more times.
1589 CURLY sv 2 Match this simple thing {n,m} times.
1590 CURLYN no 2 Match next-after-this simple thing
1591 # {n,m} times, set parenths.
1592 CURLYM no 2 Match this medium-complex thing {n,m} times.
1593 CURLYX sv 2 Match this complex thing {n,m} times.
1595 # This terminator creates a loop structure for CURLYX
1596 WHILEM no Do curly processing and see if rest matches.
1598 # OPEN,CLOSE,GROUPP ...are numbered at compile time.
1599 OPEN num 1 Mark this point in input as start of #n.
1600 CLOSE num 1 Analogous to OPEN.
1602 REF num 1 Match some already matched string
1603 REFF num 1 Match already matched string, folded
1604 REFFL num 1 Match already matched string, folded in loc.
1606 # grouping assertions
1607 IFMATCH off 1 2 Succeeds if the following matches.
1608 UNLESSM off 1 2 Fails if the following matches.
1609 SUSPEND off 1 1 "Independent" sub-RE.
1610 IFTHEN off 1 1 Switch, should be preceeded by switcher .
1611 GROUPP num 1 Whether the group matched.
1613 # Support for long RE
1614 LONGJMP off 1 1 Jump far away.
1615 BRANCHJ off 1 1 BRANCH with long offset.
1618 EVAL evl 1 Execute some Perl code.
1621 MINMOD no Next operator is not greedy.
1622 LOGICAL no Next opcode should set the flag only.
1624 # This is not used yet
1625 RENUM off 1 1 Group with independently numbered parens.
1627 # This is not really a node, but an optimized away piece of a "long" node.
1628 # To simplify debugging output, we mark it as if it were a node
1629 OPTIMIZED off Placeholder for dump.
1631 =head2 Run-time output
1633 First of all, when doing a match, one may get no run-time output even
1634 if debugging is enabled. this means that the RE engine was never
1635 entered, all of the job was done by the optimizer.
1637 If RE engine was entered, the output may look like this:
1639 Matching `[bc]d(ef*g)+h[ij]k$' against `abcdefg__gh__'
1640 Setting an EVAL scope, savestack=3
1641 2 <ab> <cdefg__gh_> | 1: ANYOF
1642 3 <abc> <defg__gh_> | 11: EXACT <d>
1643 4 <abcd> <efg__gh_> | 13: CURLYX {1,32767}
1644 4 <abcd> <efg__gh_> | 26: WHILEM
1645 0 out of 1..32767 cc=effff31c
1646 4 <abcd> <efg__gh_> | 15: OPEN1
1647 4 <abcd> <efg__gh_> | 17: EXACT <e>
1648 5 <abcde> <fg__gh_> | 19: STAR
1649 EXACT <f> can match 1 times out of 32767...
1650 Setting an EVAL scope, savestack=3
1651 6 <bcdef> <g__gh__> | 22: EXACT <g>
1652 7 <bcdefg> <__gh__> | 24: CLOSE1
1653 7 <bcdefg> <__gh__> | 26: WHILEM
1654 1 out of 1..32767 cc=effff31c
1655 Setting an EVAL scope, savestack=12
1656 7 <bcdefg> <__gh__> | 15: OPEN1
1657 7 <bcdefg> <__gh__> | 17: EXACT <e>
1658 restoring \1 to 4(4)..7
1659 failed, try continuation...
1660 7 <bcdefg> <__gh__> | 27: NOTHING
1661 7 <bcdefg> <__gh__> | 28: EXACT <h>
1665 The most significant information in the output is about the particular I<node>
1666 of the compiled RE which is currently being tested against the target string.
1667 The format of these lines is
1669 C< >I<STRING-OFFSET> <I<PRE-STRING>> <I<POST-STRING>> |I<ID>: I<TYPE>
1671 The I<TYPE> info is indented with respect to the backtracking level.
1672 Other incidental information appears interspersed within.