path: root/upstream/debian-unstable/man1/perldebguts.1

.\" -*- mode: troff; coding: utf-8 -*-
.\" Automatically generated by Pod::Man 5.01 (Pod::Simple 3.43)
.\"
.\" Standard preamble:
.\" ========================================================================
.de Sp \" Vertical space (when we can't use .PP)
.if t .sp .5v
.if n .sp
..
.de Vb \" Begin verbatim text
.ft CW
.nf
.ne \\$1
..
.de Ve \" End verbatim text
.ft R
.fi
..
.\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>.
.ie n \{\
.    ds C` ""
.    ds C' ""
'br\}
.el\{\
.    ds C`
.    ds C'
'br\}
.\"
.\" Escape single quotes in literal strings from groff's Unicode transform.
.ie \n(.g .ds Aq \(aq
.el       .ds Aq '
.\"
.\" If the F register is >0, we'll generate index entries on stderr for
.\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index
.\" entries marked with X<> in POD.  Of course, you'll have to process the
.\" output yourself in some meaningful fashion.
.\"
.\" Avoid warning from groff about undefined register 'F'.
.de IX
..
.nr rF 0
.if \n(.g .if rF .nr rF 1
.if (\n(rF:(\n(.g==0)) \{\
.    if \nF \{\
.        de IX
.        tm Index:\\$1\t\\n%\t"\\$2"
..
.        if !\nF==2 \{\
.            nr % 0
.            nr F 2
.        \}
.    \}
.\}
.rr rF
.\" ========================================================================
.\"
.IX Title "PERLDEBGUTS 1"
.TH PERLDEBGUTS 1 2024-05-30 "perl v5.38.2" "Perl Programmers Reference Guide"
.\" For nroff, turn off justification.  Always turn off hyphenation; it makes
.\" way too many mistakes in technical documents.
.if n .ad l
.nh
.SH NAME
perldebguts \- Guts of Perl debugging
.SH DESCRIPTION
.IX Header "DESCRIPTION"
This is not perldebug, which tells you how to use
the debugger.  This manpage describes low-level details concerning
the debugger's internals, which range from difficult to impossible
to understand for anyone who isn't incredibly intimate with Perl's guts.
Caveat lector.
.SH "Debugger Internals"
.IX Header "Debugger Internals"
Perl has special debugging hooks at compile-time and run-time used
to create debugging environments.  These hooks are not to be confused
with the \fIperl \-Dxxx\fR command described in perlrun,
which is usable only if a special Perl is built per the instructions in
the \fIINSTALL\fR file in the Perl source tree.
.PP
For example, whenever you call Perl's built-in \f(CW\*(C`caller\*(C'\fR function
from the package \f(CW\*(C`DB\*(C'\fR, the arguments that the corresponding stack
frame was called with are copied to the \f(CW@DB::args\fR array.  These
mechanisms are enabled by calling Perl with the \fB\-d\fR switch.
Specifically, the following additional features are enabled
(cf. "$^P" in perlvar):
.IP \(bu 4
Perl inserts the contents of \f(CW$ENV{PERL5DB}\fR (or \f(CW\*(C`BEGIN {require
\&\*(Aqperl5db.pl\*(Aq}\*(C'\fR if not present) before the first line of your program.
.IP \(bu 4
Each array \f(CW\*(C`@{"_<$filename"}\*(C'\fR holds the lines of \f(CW$filename\fR for a
file compiled by Perl.  The same is also true for \f(CW\*(C`eval\*(C'\fRed strings
that contain subroutines, or which are currently being executed.
The \f(CW$filename\fR for \f(CW\*(C`eval\*(C'\fRed strings looks like \f(CW\*(C`(eval 34)\*(C'\fR.
.Sp
Values in this array are magical in numeric context: they compare
equal to zero only if the line is not breakable.
.IP \(bu 4
Each hash \f(CW\*(C`%{"_<$filename"}\*(C'\fR contains breakpoints and actions keyed
by line number.  Individual entries (as opposed to the whole hash)
are settable.  Perl only cares about Boolean true here, although
the values used by \fIperl5db.pl\fR have the form
\&\f(CW"$break_condition\e0$action"\fR.
.Sp
The same holds for evaluated strings that contain subroutines, or
which are currently being executed.  The \f(CW$filename\fR for \f(CW\*(C`eval\*(C'\fRed strings
looks like \f(CW\*(C`(eval 34)\*(C'\fR.
.IP \(bu 4
Each scalar \f(CW\*(C`${"_<$filename"}\*(C'\fR contains \f(CW$filename\fR.  This is
also the case for evaluated strings that contain subroutines, or
which are currently being executed.  The \f(CW$filename\fR for \f(CW\*(C`eval\*(C'\fRed
strings looks like \f(CW\*(C`(eval 34)\*(C'\fR.
.IP \(bu 4
After each \f(CW\*(C`require\*(C'\fRd file is compiled, but before it is executed,
\&\f(CWDB::postponed(*{"_<$filename"})\fR is called if the subroutine
\&\f(CW\*(C`DB::postponed\*(C'\fR exists.  Here, the \f(CW$filename\fR is the expanded name of
the \f(CW\*(C`require\*(C'\fRd file, as found in the values of \f(CW%INC\fR.
.IP \(bu 4
After each subroutine \f(CW\*(C`subname\*(C'\fR is compiled, the existence of
\&\f(CW$DB::postponed{subname}\fR is checked.  If this key exists,
\&\f(CWDB::postponed(subname)\fR is called if the \f(CW\*(C`DB::postponed\*(C'\fR subroutine
also exists.
.IP \(bu 4
A hash \f(CW%DB::sub\fR is maintained, whose keys are subroutine names
and whose values have the form \f(CW\*(C`filename:startline\-endline\*(C'\fR.
\&\f(CW\*(C`filename\*(C'\fR has the form \f(CW\*(C`(eval 34)\*(C'\fR for subroutines defined inside
\&\f(CW\*(C`eval\*(C'\fRs.
.IP \(bu 4
When the execution of your program reaches a point that can hold a
breakpoint, the \f(CWDB::DB()\fR subroutine is called if any of the variables
\&\f(CW$DB::trace\fR, \f(CW$DB::single\fR, or \f(CW$DB::signal\fR is true.  These variables
are not \f(CW\*(C`local\*(C'\fRizable.  This feature is disabled when executing
inside \f(CWDB::DB()\fR, including functions called from it 
unless \f(CW\*(C`$^D & (1<<30)\*(C'\fR is true.
.IP \(bu 4
When execution of the program reaches a subroutine call, a call to
\&\f(CW&DB::sub\fR(\fIargs\fR) is made instead, with \f(CW$DB::sub\fR set to identify
the called subroutine.  (This doesn't happen if the calling subroutine
was compiled in the \f(CW\*(C`DB\*(C'\fR package.)  \f(CW$DB::sub\fR normally holds the name
of the called subroutine, if it has a name by which it can be looked up.
Failing that, \f(CW$DB::sub\fR will hold a reference to the called subroutine.
Either way, the \f(CW&DB::sub\fR subroutine can use \f(CW$DB::sub\fR as a reference
by which to call the called subroutine, which it will normally want to do.
.Sp
If the call is to an lvalue subroutine, and \f(CW&DB::lsub\fR
is defined \f(CW&DB::lsub\fR(\fIargs\fR) is called instead, otherwise falling
back to \f(CW&DB::sub\fR(\fIargs\fR).
.IX Xref "&DB::lsub"
.IP \(bu 4
When execution of the program uses \f(CW\*(C`goto\*(C'\fR to enter a non-XS subroutine
and the 0x80 bit is set in \f(CW$^P\fR, a call to \f(CW&DB::goto\fR is made, with
\&\f(CW$DB::sub\fR set to identify the subroutine being entered.  The call to
\&\f(CW&DB::goto\fR does not replace the \f(CW\*(C`goto\*(C'\fR; the requested subroutine will
still be entered once \f(CW&DB::goto\fR has returned.  \f(CW$DB::sub\fR normally
holds the name of the subroutine being entered, if it has one.  Failing
that, \f(CW$DB::sub\fR will hold a reference to the subroutine being entered.
Unlike when \f(CW&DB::sub\fR is called, it is not guaranteed that \f(CW$DB::sub\fR
can be used as a reference to operate on the subroutine being entered.
.PP
Note that if \f(CW&DB::sub\fR needs external data for it to work, no
subroutine call is possible without it. As an example, the standard
debugger's \f(CW&DB::sub\fR depends on the \f(CW$DB::deep\fR variable
(it defines how many levels of recursion deep into the debugger you can go
before a mandatory break).  If \f(CW$DB::deep\fR is not defined, subroutine
calls are not possible, even though \f(CW&DB::sub\fR exists.
.SS "Writing Your Own Debugger"
.IX Subsection "Writing Your Own Debugger"
\fIEnvironment Variables\fR
.IX Subsection "Environment Variables"
.PP
The \f(CW\*(C`PERL5DB\*(C'\fR environment variable can be used to define a debugger.
For example, the minimal "working" debugger (it actually doesn't do anything)
consists of one line:
.PP
.Vb 1
\&  sub DB::DB {}
.Ve
.PP
It can easily be defined like this:
.PP
.Vb 1
\&  $ PERL5DB="sub DB::DB {}" perl \-d your\-script
.Ve
.PP
Another brief debugger, slightly more useful, can be created
with only the line:
.PP
.Vb 1
\&  sub DB::DB {print ++$i; scalar <STDIN>}
.Ve
.PP
This debugger prints a number which increments for each statement
encountered and waits for you to hit a newline before continuing
to the next statement.
.PP
The following debugger is actually useful:
.PP
.Vb 5
\&  {
\&    package DB;
\&    sub DB  {}
\&    sub sub {print ++$i, " $sub\en"; &$sub}
\&  }
.Ve
.PP
It prints the sequence number of each subroutine call and the name of the
called subroutine.  Note that \f(CW&DB::sub\fR is being compiled into the
package \f(CW\*(C`DB\*(C'\fR through the use of the \f(CW\*(C`package\*(C'\fR directive.
.PP
When it starts, the debugger reads your rc file (\fI./.perldb\fR or
\&\fI~/.perldb\fR under Unix), which can set important options.
(A subroutine (\f(CW&afterinit\fR) can be defined here as well; it is executed
after the debugger completes its own initialization.)
.PP
After the rc file is read, the debugger reads the PERLDB_OPTS
environment variable and uses it to set debugger options. The
contents of this variable are treated as if they were the argument
of an \f(CW\*(C`o ...\*(C'\fR debugger command (q.v. in "Configurable Options" in perldebug).
.PP
\fIDebugger Internal Variables\fR
.IX Subsection "Debugger Internal Variables"
.PP
In addition to the file and subroutine-related variables mentioned above,
the debugger also maintains various magical internal variables.
.IP \(bu 4
\&\f(CW@DB::dbline\fR is an alias for \f(CW\*(C`@{"::_<current_file"}\*(C'\fR, which
holds the lines of the currently-selected file (compiled by Perl), either
explicitly chosen with the debugger's \f(CW\*(C`f\*(C'\fR command, or implicitly by flow
of execution.
.Sp
Values in this array are magical in numeric context: they compare
equal to zero only if the line is not breakable.
.IP \(bu 4
\&\f(CW%DB::dbline\fR is an alias for \f(CW\*(C`%{"::_<current_file"}\*(C'\fR, which
contains breakpoints and actions keyed by line number in
the currently-selected file, either explicitly chosen with the
debugger's \f(CW\*(C`f\*(C'\fR command, or implicitly by flow of execution.
.Sp
As previously noted, individual entries (as opposed to the whole hash)
are settable.  Perl only cares about Boolean true here, although
the values used by \fIperl5db.pl\fR have the form
\&\f(CW"$break_condition\e0$action"\fR.
.PP
\fIDebugger Customization Functions\fR
.IX Subsection "Debugger Customization Functions"
.PP
Some functions are provided to simplify customization.
.IP \(bu 4
See "Configurable Options" in perldebug for a description of options parsed by
\&\f(CWDB::parse_options(string)\fR.
.IP \(bu 4
\&\f(CW\*(C`DB::dump_trace(skip[,count])\*(C'\fR skips the specified number of frames
and returns a list containing information about the calling frames (all
of them, if \f(CW\*(C`count\*(C'\fR is missing).  Each entry is reference to a hash
with keys \f(CW\*(C`context\*(C'\fR (either \f(CW\*(C`.\*(C'\fR, \f(CW\*(C`$\*(C'\fR, or \f(CW\*(C`@\*(C'\fR), \f(CW\*(C`sub\*(C'\fR (subroutine
name, or info about \f(CW\*(C`eval\*(C'\fR), \f(CW\*(C`args\*(C'\fR (\f(CW\*(C`undef\*(C'\fR or a reference to
an array), \f(CW\*(C`file\*(C'\fR, and \f(CW\*(C`line\*(C'\fR.
.IP \(bu 4
\&\f(CW\*(C`DB::print_trace(FH, skip[, count[, short]])\*(C'\fR prints
formatted info about caller frames.  The last two functions may be
convenient as arguments to \f(CW\*(C`<\*(C'\fR, \f(CW\*(C`<<\*(C'\fR commands.
.PP
Note that any variables and functions that are not documented in
this manpages (or in perldebug) are considered for internal   
use only, and as such are subject to change without notice.
.SH "Frame Listing Output Examples"
.IX Header "Frame Listing Output Examples"
The \f(CW\*(C`frame\*(C'\fR option can be used to control the output of frame 
information.  For example, contrast this expression trace:
.PP
.Vb 2
\& $ perl \-de 42
\& Stack dump during die enabled outside of evals.
\&
\& Loading DB routines from perl5db.pl patch level 0.94
\& Emacs support available.
\&
\& Enter h or \*(Aqh h\*(Aq for help.
\&
\& main::(\-e:1):   0
\&   DB<1> sub foo { 14 }
\&
\&   DB<2> sub bar { 3 }
\&
\&   DB<3> t print foo() * bar()
\& main::((eval 172):3):   print foo() + bar();
\& main::foo((eval 168):2):
\& main::bar((eval 170):2):
\& 42
.Ve
.PP
with this one, once the \f(CW\*(C`o\*(C'\fRption \f(CW\*(C`frame=2\*(C'\fR has been set:
.PP
.Vb 11
\&   DB<4> o f=2
\&                frame = \*(Aq2\*(Aq
\&   DB<5> t print foo() * bar()
\& 3:      foo() * bar()
\& entering main::foo
\&  2:     sub foo { 14 };
\& exited main::foo
\& entering main::bar
\&  2:     sub bar { 3 };
\& exited main::bar
\& 42
.Ve
.PP
By way of demonstration, we present below a laborious listing
resulting from setting your \f(CW\*(C`PERLDB_OPTS\*(C'\fR environment variable to
the value \f(CW\*(C`f=n N\*(C'\fR, and running \fIperl \-d \-V\fR from the command line.
Examples using various values of \f(CW\*(C`n\*(C'\fR are shown to give you a feel
for the difference between settings.  Long though it may be, this
is not a complete listing, but only excerpts.
.IP 1. 4
.Vb 10
\& entering main::BEGIN
\&  entering Config::BEGIN
\&   Package lib/Exporter.pm.
\&   Package lib/Carp.pm.
\&  Package lib/Config.pm.
\&  entering Config::TIEHASH
\&  entering Exporter::import
\&   entering Exporter::export
\& entering Config::myconfig
\&  entering Config::FETCH
\&  entering Config::FETCH
\&  entering Config::FETCH
\&  entering Config::FETCH
.Ve
.IP 2. 4
.Vb 10
\& entering main::BEGIN
\&  entering Config::BEGIN
\&   Package lib/Exporter.pm.
\&   Package lib/Carp.pm.
\&  exited Config::BEGIN
\&  Package lib/Config.pm.
\&  entering Config::TIEHASH
\&  exited Config::TIEHASH
\&  entering Exporter::import
\&   entering Exporter::export
\&   exited Exporter::export
\&  exited Exporter::import
\& exited main::BEGIN
\& entering Config::myconfig
\&  entering Config::FETCH
\&  exited Config::FETCH
\&  entering Config::FETCH
\&  exited Config::FETCH
\&  entering Config::FETCH
.Ve
.IP 3. 4
.Vb 10
\& in  $=main::BEGIN() from /dev/null:0
\&  in  $=Config::BEGIN() from lib/Config.pm:2
\&   Package lib/Exporter.pm.
\&   Package lib/Carp.pm.
\&  Package lib/Config.pm.
\&  in  $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:644
\&  in  $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\&   in  $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from li
\& in  @=Config::myconfig() from /dev/null:0
\&  in  $=Config::FETCH(ref(Config), \*(Aqpackage\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(Aqbaserev\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(AqPERL_VERSION\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(AqPERL_SUBVERSION\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(Aqosname\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(Aqosvers\*(Aq) from lib/Config.pm:574
.Ve
.IP 4. 4
.Vb 10
\& in  $=main::BEGIN() from /dev/null:0
\&  in  $=Config::BEGIN() from lib/Config.pm:2
\&   Package lib/Exporter.pm.
\&   Package lib/Carp.pm.
\&  out $=Config::BEGIN() from lib/Config.pm:0
\&  Package lib/Config.pm.
\&  in  $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:644
\&  out $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:644
\&  in  $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\&   in  $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/
\&   out $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/
\&  out $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\& out $=main::BEGIN() from /dev/null:0
\& in  @=Config::myconfig() from /dev/null:0
\&  in  $=Config::FETCH(ref(Config), \*(Aqpackage\*(Aq) from lib/Config.pm:574
\&  out $=Config::FETCH(ref(Config), \*(Aqpackage\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(Aqbaserev\*(Aq) from lib/Config.pm:574
\&  out $=Config::FETCH(ref(Config), \*(Aqbaserev\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(AqPERL_VERSION\*(Aq) from lib/Config.pm:574
\&  out $=Config::FETCH(ref(Config), \*(AqPERL_VERSION\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(ref(Config), \*(AqPERL_SUBVERSION\*(Aq) from lib/Config.pm:574
.Ve
.IP 5. 4
.Vb 10
\& in  $=main::BEGIN() from /dev/null:0
\&  in  $=Config::BEGIN() from lib/Config.pm:2
\&   Package lib/Exporter.pm.
\&   Package lib/Carp.pm.
\&  out $=Config::BEGIN() from lib/Config.pm:0
\&  Package lib/Config.pm.
\&  in  $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:644
\&  out $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:644
\&  in  $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\&   in  $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/E
\&   out $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/E
\&  out $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\& out $=main::BEGIN() from /dev/null:0
\& in  @=Config::myconfig() from /dev/null:0
\&  in  $=Config::FETCH(\*(AqConfig=HASH(0x1aa444)\*(Aq, \*(Aqpackage\*(Aq) from lib/Config.pm:574
\&  out $=Config::FETCH(\*(AqConfig=HASH(0x1aa444)\*(Aq, \*(Aqpackage\*(Aq) from lib/Config.pm:574
\&  in  $=Config::FETCH(\*(AqConfig=HASH(0x1aa444)\*(Aq, \*(Aqbaserev\*(Aq) from lib/Config.pm:574
\&  out $=Config::FETCH(\*(AqConfig=HASH(0x1aa444)\*(Aq, \*(Aqbaserev\*(Aq) from lib/Config.pm:574
.Ve
.IP 6. 4
.Vb 10
\& in  $=CODE(0x15eca4)() from /dev/null:0
\&  in  $=CODE(0x182528)() from lib/Config.pm:2
\&   Package lib/Exporter.pm.
\&  out $=CODE(0x182528)() from lib/Config.pm:0
\&  scalar context return from CODE(0x182528): undef
\&  Package lib/Config.pm.
\&  in  $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:628
\&  out $=Config::TIEHASH(\*(AqConfig\*(Aq) from lib/Config.pm:628
\&  scalar context return from Config::TIEHASH:   empty hash
\&  in  $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\&   in  $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/Exporter.pm:171
\&   out $=Exporter::export(\*(AqConfig\*(Aq, \*(Aqmain\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from lib/Exporter.pm:171
\&   scalar context return from Exporter::export: \*(Aq\*(Aq
\&  out $=Exporter::import(\*(AqConfig\*(Aq, \*(Aqmyconfig\*(Aq, \*(Aqconfig_vars\*(Aq) from /dev/null:0
\&  scalar context return from Exporter::import: \*(Aq\*(Aq
.Ve
.PP
In all cases shown above, the line indentation shows the call tree.
If bit 2 of \f(CW\*(C`frame\*(C'\fR is set, a line is printed on exit from a
subroutine as well.  If bit 4 is set, the arguments are printed
along with the caller info.  If bit 8 is set, the arguments are
printed even if they are tied or references.  If bit 16 is set, the
return value is printed, too.
.PP
When a package is compiled, a line like this
.PP
.Vb 1
\&    Package lib/Carp.pm.
.Ve
.PP
is printed with proper indentation.
.SH "Debugging Regular Expressions"
.IX Header "Debugging Regular Expressions"
There are two ways to enable debugging output for regular expressions.
.PP
If your perl is compiled with \f(CW\*(C`\-DDEBUGGING\*(C'\fR, you may use the
\&\fB\-Dr\fR flag on the command line, and \f(CW\*(C`\-Drv\*(C'\fR for more verbose
information.
.PP
Otherwise, one can \f(CW\*(C`use re \*(Aqdebug\*(Aq\*(C'\fR, which has effects at both
compile time and run time.  Since Perl 5.9.5, this pragma is lexically
scoped.
.SS "Compile-time Output"
.IX Subsection "Compile-time Output"
The debugging output at compile time looks like this:
.PP
.Vb 10
\&  Compiling REx \*(Aq[bc]d(ef*g)+h[ij]k$\*(Aq
\&  size 45 Got 364 bytes for offset annotations.
\&  first at 1
\&  rarest char g at 0
\&  rarest char d at 0
\&     1: ANYOF[bc](12)
\&    12: EXACT <d>(14)
\&    14: CURLYX[0] {1,32767}(28)
\&    16:   OPEN1(18)
\&    18:     EXACT <e>(20)
\&    20:     STAR(23)
\&    21:       EXACT <f>(0)
\&    23:     EXACT <g>(25)
\&    25:   CLOSE1(27)
\&    27:   WHILEM[1/1](0)
\&    28: NOTHING(29)
\&    29: EXACT <h>(31)
\&    31: ANYOF[ij](42)
\&    42: EXACT <k>(44)
\&    44: EOL(45)
\&    45: END(0)
\&  anchored \*(Aqde\*(Aq at 1 floating \*(Aqgh\*(Aq at 3..2147483647 (checking floating) 
\&        stclass \*(AqANYOF[bc]\*(Aq minlen 7 
\&  Offsets: [45]
\&        1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
\&        0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
\&        11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
\&        0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]  
\&  Omitting $\` $& $\*(Aq support.
.Ve
.PP
The first line shows the pre-compiled form of the regex.  The second
shows the size of the compiled form (in arbitrary units, usually
4\-byte words) and the total number of bytes allocated for the
offset/length table, usually 4+\f(CW\*(C`size\*(C'\fR*8.  The next line shows the
label \fIid\fR of the first node that does a match.
.PP
The
.PP
.Vb 2
\&  anchored \*(Aqde\*(Aq at 1 floating \*(Aqgh\*(Aq at 3..2147483647 (checking floating) 
\&        stclass \*(AqANYOF[bc]\*(Aq minlen 7
.Ve
.PP
line (split into two lines above) contains optimizer
information.  In the example shown, the optimizer found that the match 
should contain a substring \f(CW\*(C`de\*(C'\fR at offset 1, plus substring \f(CW\*(C`gh\*(C'\fR
at some offset between 3 and infinity.  Moreover, when checking for
these substrings (to abandon impossible matches quickly), Perl will check
for the substring \f(CW\*(C`gh\*(C'\fR before checking for the substring \f(CW\*(C`de\*(C'\fR.  The
optimizer may also use the knowledge that the match starts (at the
\&\f(CW\*(C`first\*(C'\fR \fIid\fR) with a character class, and no string 
shorter than 7 characters can possibly match.
.PP
The fields of interest which may appear in this line are
.ie n .IP """anchored"" \fISTRING\fR ""at"" \fIPOS\fR" 4
.el .IP "\f(CWanchored\fR \fISTRING\fR \f(CWat\fR \fIPOS\fR" 4
.IX Item "anchored STRING at POS"
.PD 0
.ie n .IP """floating"" \fISTRING\fR ""at"" \fIPOS1..POS2\fR" 4
.el .IP "\f(CWfloating\fR \fISTRING\fR \f(CWat\fR \fIPOS1..POS2\fR" 4
.IX Item "floating STRING at POS1..POS2"
.PD
See above.
.ie n .IP """matching floating/anchored""" 4
.el .IP "\f(CWmatching floating/anchored\fR" 4
.IX Item "matching floating/anchored"
Which substring to check first.
.ie n .IP """minlen""" 4
.el .IP \f(CWminlen\fR 4
.IX Item "minlen"
The minimal length of the match.
.ie n .IP """stclass"" \fITYPE\fR" 4
.el .IP "\f(CWstclass\fR \fITYPE\fR" 4
.IX Item "stclass TYPE"
Type of first matching node.
.ie n .IP """noscan""" 4
.el .IP \f(CWnoscan\fR 4
.IX Item "noscan"
Don't scan for the found substrings.
.ie n .IP """isall""" 4
.el .IP \f(CWisall\fR 4
.IX Item "isall"
Means that the optimizer information is all that the regular
expression contains, and thus one does not need to enter the regex engine at
all.
.ie n .IP """GPOS""" 4
.el .IP \f(CWGPOS\fR 4
.IX Item "GPOS"
Set if the pattern contains \f(CW\*(C`\eG\*(C'\fR.
.ie n .IP """plus""" 4
.el .IP \f(CWplus\fR 4
.IX Item "plus"
Set if the pattern starts with a repeated char (as in \f(CW\*(C`x+y\*(C'\fR).
.ie n .IP """implicit""" 4
.el .IP \f(CWimplicit\fR 4
.IX Item "implicit"
Set if the pattern starts with \f(CW\*(C`.*\*(C'\fR.
.ie n .IP """with eval""" 4
.el .IP "\f(CWwith eval\fR" 4
.IX Item "with eval"
Set if the pattern contain eval-groups, such as \f(CW\*(C`(?{ code })\*(C'\fR and
\&\f(CW\*(C`(??{ code })\*(C'\fR.
.ie n .IP anchored(TYPE) 4
.el .IP \f(CWanchored(TYPE)\fR 4
.IX Item "anchored(TYPE)"
If the pattern may match only at a handful of places, with \f(CW\*(C`TYPE\*(C'\fR
being \f(CW\*(C`SBOL\*(C'\fR, \f(CW\*(C`MBOL\*(C'\fR, or \f(CW\*(C`GPOS\*(C'\fR.  See the table below.
.PP
If a substring is known to match at end-of-line only, it may be
followed by \f(CW\*(C`$\*(C'\fR, as in \f(CW\*(C`floating \*(Aqk\*(Aq$\*(C'\fR.
.PP
The optimizer-specific information is used to avoid entering (a slow) regex
engine on strings that will not definitely match.  If the \f(CW\*(C`isall\*(C'\fR flag
is set, a call to the regex engine may be avoided even when the optimizer
found an appropriate place for the match.
.PP
Above the optimizer section is the list of \fInodes\fR of the compiled
form of the regex.  Each line has format
.PP
\&\f(CW\*(C`   \*(C'\fR\fIid\fR: \fITYPE\fR \fIOPTIONAL-INFO\fR (\fInext-id\fR)
.SS "Types of Nodes"
.IX Subsection "Types of Nodes"
Here are the current possible types, with short descriptions:
.PP
.Vb 1
\& # TYPE arg\-description [regnode\-struct\-suffix] [longjump\-len] DESCRIPTION
\&
\& # Exit points
\&
\& END              no         End of program.
\& SUCCEED          no         Return from a subroutine, basically.
\&
\& # Line Start Anchors:
\& SBOL             no         Match "" at beginning of line: /^/, /\eA/
\& MBOL             no         Same, assuming multiline: /^/m
\&
\& # Line End Anchors:
\& SEOL             no         Match "" at end of line: /$/
\& MEOL             no         Same, assuming multiline: /$/m
\& EOS              no         Match "" at end of string: /\ez/
\&
\& # Match Start Anchors:
\& GPOS             no         Matches where last m//g left off.
\&
\& # Word Boundary Opcodes:
\& BOUND            no         Like BOUNDA for non\-utf8, otherwise like
\&                             BOUNDU
\& BOUNDL           no         Like BOUND/BOUNDU, but \ew and \eW are
\&                             defined by current locale
\& BOUNDU           no         Match "" at any boundary of a given type
\&                             using /u rules.
\& BOUNDA           no         Match "" at any boundary between \ew\eW or
\&                             \eW\ew, where \ew is [_a\-zA\-Z0\-9]
\& NBOUND           no         Like NBOUNDA for non\-utf8, otherwise like
\&                             BOUNDU
\& NBOUNDL          no         Like NBOUND/NBOUNDU, but \ew and \eW are
\&                             defined by current locale
\& NBOUNDU          no         Match "" at any non\-boundary of a given
\&                             type using using /u rules.
\& NBOUNDA          no         Match "" betweeen any \ew\ew or \eW\eW, where
\&                             \ew is [_a\-zA\-Z0\-9]
\&
\& # [Special] alternatives:
\& REG_ANY          no         Match any one character (except newline).
\& SANY             no         Match any one character.
\& ANYOF            sv         Match character in (or not in) this class,
\&                  charclass  single char match only
\& ANYOFD           sv         Like ANYOF, but /d is in effect
\&                  charclass
\& ANYOFL           sv         Like ANYOF, but /l is in effect
\&                  charclass
\& ANYOFPOSIXL      sv         Like ANYOFL, but matches [[:posix:]]
\&                  charclass_ classes
\&                  posixl
\&
\& ANYOFH           sv 1       Like ANYOF, but only has "High" matches,
\&                             none in the bitmap; the flags field
\&                             contains the lowest matchable UTF\-8 start
\&                             byte
\& ANYOFHb          sv 1       Like ANYOFH, but all matches share the same
\&                             UTF\-8 start byte, given in the flags field
\& ANYOFHr          sv 1       Like ANYOFH, but the flags field contains
\&                             packed bounds for all matchable UTF\-8 start
\&                             bytes.
\& ANYOFHs          sv:str 1   Like ANYOFHb, but has a string field that
\&                             gives the leading matchable UTF\-8 bytes;
\&                             flags field is len
\& ANYOFR           packed 1   Matches any character in the range given by
\&                             its packed args: upper 12 bits is the max
\&                             delta from the base lower 20; the flags
\&                             field contains the lowest matchable UTF\-8
\&                             start byte
\& ANYOFRb          packed 1   Like ANYOFR, but all matches share the same
\&                             UTF\-8 start byte, given in the flags field
\&
\& ANYOFHbbm        none bbm   Like ANYOFHb, but only for 2\-byte UTF\-8
\&                             characters; uses a bitmap to match the
\&                             continuation byte
\&
\& ANYOFM           byte 1     Like ANYOF, but matches an invariant byte
\&                             as determined by the mask and arg
\& NANYOFM          byte 1     complement of ANYOFM
\&
\& # POSIX Character Classes:
\& POSIXD           none       Some [[:class:]] under /d; the FLAGS field
\&                             gives which one
\& POSIXL           none       Some [[:class:]] under /l; the FLAGS field
\&                             gives which one
\& POSIXU           none       Some [[:class:]] under /u; the FLAGS field
\&                             gives which one
\& POSIXA           none       Some [[:class:]] under /a; the FLAGS field
\&                             gives which one
\& NPOSIXD          none       complement of POSIXD, [[:^class:]]
\& NPOSIXL          none       complement of POSIXL, [[:^class:]]
\& NPOSIXU          none       complement of POSIXU, [[:^class:]]
\& NPOSIXA          none       complement of POSIXA, [[:^class:]]
\&
\& CLUMP            no         Match any extended grapheme cluster
\&                             sequence
\&
\& # Alternation
\&
\& # BRANCH        The set of branches constituting a single choice are
\& #               hooked together with their "next" pointers, since
\& #               precedence prevents anything being concatenated to
\& #               any individual branch.  The "next" pointer of the last
\& #               BRANCH in a choice points to the thing following the
\& #               whole choice.  This is also where the final "next"
\& #               pointer of each individual branch points; each branch
\& #               starts with the operand node of a BRANCH node.
\& #
\& BRANCH           node 1     Match this alternative, or the next...
\&
\& # Literals
\&
\& EXACT            str        Match this string (flags field is the
\&                             length).
\&
\& # In a long string node, the U32 argument is the length, and is
\& # immediately followed by the string.
\& LEXACT           len:str 1  Match this long string (preceded by length;
\&                             flags unused).
\& EXACTL           str        Like EXACT, but /l is in effect (used so
\&                             locale\-related warnings can be checked for)
\& EXACTF           str        Like EXACT, but match using /id rules;
\&                             (string not UTF\-8, ASCII folded; non\-ASCII
\&                             not)
\& EXACTFL          str        Like EXACT, but match using /il rules;
\&                             (string not likely to be folded)
\& EXACTFU          str        Like EXACT, but match using /iu rules;
\&                             (string folded)
\&
\& EXACTFAA         str        Like EXACT, but match using /iaa rules;
\&                             (string folded except MICRO in non\-UTF8
\&                             patterns; doesn\*(Aqt contain SHARP S unless
\&                             UTF\-8; folded length <= unfolded)
\& EXACTFAA_NO_TRIE str        Like EXACTFAA, (string not UTF\-8, folded
\&                             except: MICRO, SHARP S; folded length <=
\&                             unfolded, not currently trie\-able)
\&
\& EXACTFUP         str        Like EXACT, but match using /iu rules;
\&                             (string not UTF\-8, folded except MICRO:
\&                             hence Problematic)
\&
\& EXACTFLU8        str        Like EXACTFU, but use /il, UTF\-8, (string
\&                             is folded, and everything in it is above
\&                             255
\& EXACT_REQ8       str        Like EXACT, but only UTF\-8 encoded targets
\&                             can match
\& LEXACT_REQ8      len:str 1  Like LEXACT, but only UTF\-8 encoded targets
\&                             can match
\& EXACTFU_REQ8     str        Like EXACTFU, but only UTF\-8 encoded
\&                             targets can match
\&
\& EXACTFU_S_EDGE   str        /di rules, but nothing in it precludes /ui,
\&                             except begins and/or ends with [Ss];
\&                             (string not UTF\-8; compile\-time only)
\&
\& # New charclass like patterns
\& LNBREAK          none       generic newline pattern
\&
\& # Trie Related
\&
\& # Behave the same as A|LIST|OF|WORDS would. The \*(Aq..C\*(Aq variants
\& # have inline charclass data (ascii only), the \*(AqC\*(Aq store it in the
\& # structure.
\&
\& TRIE             trie 1     Match many EXACT(F[ALU]?)? at once.
\&                             flags==type
\& TRIEC            trie       Same as TRIE, but with embedded charclass
\&                  charclass  data
\&
\& AHOCORASICK      trie 1     Aho Corasick stclass. flags==type
\& AHOCORASICKC     trie       Same as AHOCORASICK, but with embedded
\&                  charclass  charclass data
\&
\& # Do nothing types
\&
\& NOTHING          no         Match empty string.
\& # A variant of above which delimits a group, thus stops optimizations
\& TAIL             no         Match empty string. Can jump here from
\&                             outside.
\&
\& # Loops
\&
\& # STAR,PLUS    \*(Aq?\*(Aq, and complex \*(Aq*\*(Aq and \*(Aq+\*(Aq, are implemented as
\& #               circular BRANCH structures.  Simple cases
\& #               (one character per match) are implemented with STAR
\& #               and PLUS for speed and to minimize recursive plunges.
\& #
\& STAR             node       Match this (simple) thing 0 or more times:
\&                             /A{0,}B/ where A is width 1 char
\& PLUS             node       Match this (simple) thing 1 or more times:
\&                             /A{1,}B/ where A is width 1 char
\&
\& CURLY            sv 3       Match this (simple) thing {n,m} times:
\&                             /A{m,n}B/ where A is width 1 char
\& CURLYN           no 3       Capture next\-after\-this simple thing:
\&                             /(A){m,n}B/ where A is width 1 char
\& CURLYM           no 3       Capture this medium\-complex thing {n,m}
\&                             times: /(A){m,n}B/ where A is fixed\-length
\& CURLYX           sv 3       Match/Capture this complex thing {n,m}
\&                             times.
\&
\& # This terminator creates a loop structure for CURLYX
\& WHILEM           no         Do curly processing and see if rest
\&                             matches.
\&
\& # Buffer related
\&
\& # OPEN,CLOSE,GROUPP     ...are numbered at compile time.
\& OPEN             num 1      Mark this point in input as start of #n.
\& CLOSE            num 1      Close corresponding OPEN of #n.
\& SROPEN           none       Same as OPEN, but for script run
\& SRCLOSE          none       Close preceding SROPEN
\&
\& REF              num 2      Match some already matched string
\& REFF             num 2      Match already matched string, using /di
\&                             rules.
\& REFFL            num 2      Match already matched string, using /li
\&                             rules.
\& REFFU            num 2      Match already matched string, usng /ui.
\& REFFA            num 2      Match already matched string, using /aai
\&                             rules.
\&
\& # Named references.  Code in regcomp.c assumes that these all are after
\& # the numbered references
\& REFN             no\-sv 2    Match some already matched string
\& REFFN            no\-sv 2    Match already matched string, using /di
\&                             rules.
\& REFFLN           no\-sv 2    Match already matched string, using /li
\&                             rules.
\& REFFUN           num 2      Match already matched string, using /ui
\&                             rules.
\& REFFAN           num 2      Match already matched string, using /aai
\&                             rules.
\&
\& # Support for long RE
\& LONGJMP          off 1 1    Jump far away.
\& BRANCHJ          off 2 1    BRANCH with long offset.
\&
\& # Special Case Regops
\& IFMATCH          off 1 1    Succeeds if the following matches; non\-zero
\&                             flags "f", next_off "o" means lookbehind
\&                             assertion starting "f..(f\-o)" characters
\&                             before current
\& UNLESSM          off 1 1    Fails if the following matches; non\-zero
\&                             flags "f", next_off "o" means lookbehind
\&                             assertion starting "f..(f\-o)" characters
\&                             before current
\& SUSPEND          off 1 1    "Independent" sub\-RE.
\& IFTHEN           off 1 1    Switch, should be preceded by switcher.
\& GROUPP           num 1      Whether the group matched.
\&
\& # The heavy worker
\&
\& EVAL             evl/flags  Execute some Perl code.
\&                  2
\&
\& # Modifiers
\&
\& MINMOD           no         Next operator is not greedy.
\& LOGICAL          no         Next opcode should set the flag only.
\&
\& # This is not used yet
\& RENUM            off 1 1    Group with independently numbered parens.
\&
\& # Regex Subroutines
\& GOSUB            num/ofs 2  recurse to paren arg1 at (signed) ofs arg2
\&
\& # Special conditionals
\& GROUPPN          no\-sv 1    Whether the group matched.
\& INSUBP           num 1      Whether we are in a specific recurse.
\& DEFINEP          none 1     Never execute directly.
\&
\& # Backtracking Verbs
\& ENDLIKE          none       Used only for the type field of verbs
\& OPFAIL           no\-sv 1    Same as (?!), but with verb arg
\& ACCEPT           no\-sv/num  Accepts the current matched string, with
\&                  2          verbar
\&
\& # Verbs With Arguments
\& VERB             no\-sv 1    Used only for the type field of verbs
\& PRUNE            no\-sv 1    Pattern fails at this startpoint if no\-
\&                             backtracking through this
\& MARKPOINT        no\-sv 1    Push the current location for rollback by
\&                             cut.
\& SKIP             no\-sv 1    On failure skip forward (to the mark)
\&                             before retrying
\& COMMIT           no\-sv 1    Pattern fails outright if backtracking
\&                             through this
\& CUTGROUP         no\-sv 1    On failure go to the next alternation in
\&                             the group
\&
\& # Control what to keep in $&.
\& KEEPS            no         $& begins here.
\&
\& # Validate that lookbehind IFMATCH and UNLESSM end at the right place
\& LOOKBEHIND_END   no         Return from lookbehind (IFMATCH/UNLESSM)
\&                             and validate position
\&
\& # SPECIAL  REGOPS
\&
\& # This is not really a node, but an optimized away piece of a "long"
\& # node.  To simplify debugging output, we mark it as if it were a node
\& OPTIMIZED        off        Placeholder for dump.
\&
\& # Special opcode with the property that no opcode in a compiled program
\& # will ever be of this type. Thus it can be used as a flag value that
\& # no other opcode has been seen. END is used similarly, in that an END
\& # node cant be optimized. So END implies "unoptimizable" and PSEUDO
\& # mean "not seen anything to optimize yet".
\& PSEUDO           off        Pseudo opcode for internal use.
\&
\& REGEX_SET        depth p    Regex set, temporary node used in pre\-
\&                             optimization compilation
.Ve
.PP
Following the optimizer information is a dump of the offset/length
table, here split across several lines:
.PP
.Vb 5
\&  Offsets: [45]
\&        1[4] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 5[1]
\&        0[0] 12[1] 0[0] 6[1] 0[0] 7[1] 0[0] 9[1] 8[1] 0[0] 10[1] 0[0]
\&        11[1] 0[0] 12[0] 12[0] 13[1] 0[0] 14[4] 0[0] 0[0] 0[0] 0[0]
\&        0[0] 0[0] 0[0] 0[0] 0[0] 0[0] 18[1] 0[0] 19[1] 20[0]
.Ve
.PP
The first line here indicates that the offset/length table contains 45
entries.  Each entry is a pair of integers, denoted by \f(CW\*(C`offset[length]\*(C'\fR.
Entries are numbered starting with 1, so entry #1 here is \f(CW\*(C`1[4]\*(C'\fR and
entry #12 is \f(CW\*(C`5[1]\*(C'\fR.  \f(CW\*(C`1[4]\*(C'\fR indicates that the node labeled \f(CW\*(C`1:\*(C'\fR
(the \f(CW\*(C`1: ANYOF[bc]\*(C'\fR) begins at character position 1 in the
pre-compiled form of the regex, and has a length of 4 characters.
\&\f(CW\*(C`5[1]\*(C'\fR in position 12 
indicates that the node labeled \f(CW\*(C`12:\*(C'\fR
(the \f(CW\*(C`12: EXACT <d>\*(C'\fR) begins at character position 5 in the
pre-compiled form of the regex, and has a length of 1 character.
\&\f(CW\*(C`12[1]\*(C'\fR in position 14 
indicates that the node labeled \f(CW\*(C`14:\*(C'\fR
(the \f(CW\*(C`14: CURLYX[0] {1,32767}\*(C'\fR) begins at character position 12 in the
pre-compiled form of the regex, and has a length of 1 character\-\-\-that
is, it corresponds to the \f(CW\*(C`+\*(C'\fR symbol in the precompiled regex.
.PP
\&\f(CW\*(C`0[0]\*(C'\fR items indicate that there is no corresponding node.
.SS "Run-time Output"
.IX Subsection "Run-time Output"
First of all, when doing a match, one may get no run-time output even
if debugging is enabled.  This means that the regex engine was never
entered and that all of the job was therefore done by the optimizer.
.PP
If the regex engine was entered, the output may look like this:
.PP
.Vb 10
\&  Matching \*(Aq[bc]d(ef*g)+h[ij]k$\*(Aq against \*(Aqabcdefg_\|_gh_\|_\*(Aq
\&    Setting an EVAL scope, savestack=3
\&     2 <ab> <cdefg_\|_gh_>    |  1: ANYOF
\&     3 <abc> <defg_\|_gh_>    | 11: EXACT <d>
\&     4 <abcd> <efg_\|_gh_>    | 13: CURLYX {1,32767}
\&     4 <abcd> <efg_\|_gh_>    | 26:   WHILEM
\&                                0 out of 1..32767  cc=effff31c
\&     4 <abcd> <efg_\|_gh_>    | 15:     OPEN1
\&     4 <abcd> <efg_\|_gh_>    | 17:     EXACT <e>
\&     5 <abcde> <fg_\|_gh_>    | 19:     STAR
\&                             EXACT <f> can match 1 times out of 32767...
\&    Setting an EVAL scope, savestack=3
\&     6 <bcdef> <g_\|_gh_\|_>    | 22:       EXACT <g>
\&     7 <bcdefg> <_\|_gh_\|_>    | 24:       CLOSE1
\&     7 <bcdefg> <_\|_gh_\|_>    | 26:       WHILEM
\&                                    1 out of 1..32767  cc=effff31c
\&    Setting an EVAL scope, savestack=12
\&     7 <bcdefg> <_\|_gh_\|_>    | 15:         OPEN1
\&     7 <bcdefg> <_\|_gh_\|_>    | 17:         EXACT <e>
\&       restoring \e1 to 4(4)..7
\&                                    failed, try continuation...
\&     7 <bcdefg> <_\|_gh_\|_>    | 27:         NOTHING
\&     7 <bcdefg> <_\|_gh_\|_>    | 28:         EXACT <h>
\&                                    failed...
\&                                failed...
.Ve
.PP
The most significant information in the output is about the particular \fInode\fR
of the compiled regex that is currently being tested against the target string.
The format of these lines is
.PP
\&\f(CW\*(C`    \*(C'\fR\fISTRING-OFFSET\fR <\fIPRE-STRING\fR> <\fIPOST-STRING\fR>   |\fIID\fR:  \fITYPE\fR
.PP
The \fITYPE\fR info is indented with respect to the backtracking level.
Other incidental information appears interspersed within.
.SH "Debugging Perl Memory Usage"
.IX Header "Debugging Perl Memory Usage"
Perl is a profligate wastrel when it comes to memory use.  There
is a saying that to estimate memory usage of Perl, assume a reasonable
algorithm for memory allocation, multiply that estimate by 10, and
while you still may miss the mark, at least you won't be quite so
astonished.  This is not absolutely true, but may provide a good
grasp of what happens.
.PP
Assume that an integer cannot take less than 20 bytes of memory, a
float cannot take less than 24 bytes, a string cannot take less
than 32 bytes (all these examples assume 32\-bit architectures, the
result are quite a bit worse on 64\-bit architectures).  If a variable
is accessed in two of three different ways (which require an integer,
a float, or a string), the memory footprint may increase yet another
20 bytes.  A sloppy \fBmalloc\fR\|(3) implementation can inflate these
numbers dramatically.
.PP
On the opposite end of the scale, a declaration like
.PP
.Vb 1
\&  sub foo;
.Ve
.PP
may take up to 500 bytes of memory, depending on which release of Perl
you're running.
.PP
Anecdotal estimates of source-to-compiled code bloat suggest an
eightfold increase.  This means that the compiled form of reasonable
(normally commented, properly indented etc.) code will take
about eight times more space in memory than the code took
on disk.
.PP
The \fB\-DL\fR command-line switch is obsolete since circa Perl 5.6.0
(it was available only if Perl was built with \f(CW\*(C`\-DDEBUGGING\*(C'\fR).
The switch was used to track Perl's memory allocations and possible
memory leaks.  These days the use of malloc debugging tools like
\&\fIPurify\fR or \fIvalgrind\fR is suggested instead.  See also
"PERL_MEM_LOG" in perlhacktips.
.PP
One way to find out how much memory is being used by Perl data
structures is to install the Devel::Size module from CPAN: it gives
you the minimum number of bytes required to store a particular data
structure.  Please be mindful of the difference between the \fBsize()\fR
and \fBtotal_size()\fR.
.PP
If Perl has been compiled using Perl's malloc you can analyze Perl
memory usage by setting \f(CW$ENV\fR{PERL_DEBUG_MSTATS}.
.ie n .SS "Using $ENV{PERL_DEBUG_MSTATS}"
.el .SS "Using \f(CW$ENV{PERL_DEBUG_MSTATS}\fP"
.IX Subsection "Using $ENV{PERL_DEBUG_MSTATS}"
If your perl is using Perl's \fBmalloc()\fR and was compiled with the
necessary switches (this is the default), then it will print memory
usage statistics after compiling your code when \f(CW\*(C`$ENV{PERL_DEBUG_MSTATS}
> 1\*(C'\fR, and before termination of the program when \f(CW\*(C`$ENV{PERL_DEBUG_MSTATS} >= 1\*(C'\fR.  The report format is similar to
the following example:
.PP
.Vb 10
\& $ PERL_DEBUG_MSTATS=2 perl \-e "require Carp"
\& Memory allocation statistics after compilation: (buckets 4(4)..8188(8192)
\&    14216 free:   130   117    28     7     9   0   2     2   1 0 0
\&                437    61    36     0     5
\&    60924 used:   125   137   161    55     7   8   6    16   2 0 1
\&                 74   109   304    84    20
\& Total sbrk(): 77824/21:119. Odd ends: pad+heads+chain+tail: 0+636+0+2048.
\& Memory allocation statistics after execution:   (buckets 4(4)..8188(8192)
\&    30888 free:   245    78    85    13     6   2   1     3   2 0 1
\&                315   162    39    42    11
\&   175816 used:   265   176  1112   111    26  22  11    27   2 1 1
\&                196   178  1066   798    39
\& Total sbrk(): 215040/47:145. Odd ends: pad+heads+chain+tail: 0+2192+0+6144.
.Ve
.PP
It is possible to ask for such a statistic at arbitrary points in
your execution using the \fBmstat()\fR function out of the standard
Devel::Peek module.
.PP
Here is some explanation of that format:
.ie n .IP """buckets SMALLEST(APPROX)..GREATEST(APPROX)""" 4
.el .IP "\f(CWbuckets SMALLEST(APPROX)..GREATEST(APPROX)\fR" 4
.IX Item "buckets SMALLEST(APPROX)..GREATEST(APPROX)"
Perl's \fBmalloc()\fR uses bucketed allocations.  Every request is rounded
up to the closest bucket size available, and a bucket is taken from
the pool of buckets of that size.
.Sp
The line above describes the limits of buckets currently in use.
Each bucket has two sizes: memory footprint and the maximal size
of user data that can fit into this bucket.  Suppose in the above
example that the smallest bucket were size 4.  The biggest bucket
would have usable size 8188, and the memory footprint would be 8192.
.Sp
In a Perl built for debugging, some buckets may have negative usable
size.  This means that these buckets cannot (and will not) be used.
For larger buckets, the memory footprint may be one page greater
than a power of 2.  If so, the corresponding power of two is
printed in the \f(CW\*(C`APPROX\*(C'\fR field above.
.IP Free/Used 4
.IX Item "Free/Used"
The 1 or 2 rows of numbers following that correspond to the number
of buckets of each size between \f(CW\*(C`SMALLEST\*(C'\fR and \f(CW\*(C`GREATEST\*(C'\fR.  In
the first row, the sizes (memory footprints) of buckets are powers
of two\-\-or possibly one page greater.  In the second row, if present,
the memory footprints of the buckets are between the memory footprints
of two buckets "above".
.Sp
For example, suppose under the previous example, the memory footprints
were
.Sp
.Vb 2
\&   free:    8     16    32    64    128  256 512 1024 2048 4096 8192
\&           4     12    24    48    80
.Ve
.Sp
With a non\-\f(CW\*(C`DEBUGGING\*(C'\fR perl, the buckets starting from \f(CW128\fR have
a 4\-byte overhead, and thus an 8192\-long bucket may take up to
8188\-byte allocations.
.ie n .IP """Total sbrk(): SBRKed/SBRKs:CONTINUOUS""" 4
.el .IP "\f(CWTotal sbrk(): SBRKed/SBRKs:CONTINUOUS\fR" 4
.IX Item "Total sbrk(): SBRKed/SBRKs:CONTINUOUS"
The first two fields give the total amount of memory perl \fBsbrk\fR\|(2)ed
(ess-broken? :\-) and number of \fBsbrk\fR\|(2)s used.  The third number is
what perl thinks about continuity of returned chunks.  So long as
this number is positive, \fBmalloc()\fR will assume that it is probable
that \fBsbrk\fR\|(2) will provide continuous memory.
.Sp
Memory allocated by external libraries is not counted.
.ie n .IP """pad: 0""" 4
.el .IP "\f(CWpad: 0\fR" 4
.IX Item "pad: 0"
The amount of \fBsbrk\fR\|(2)ed memory needed to keep buckets aligned.
.ie n .IP """heads: 2192""" 4
.el .IP "\f(CWheads: 2192\fR" 4
.IX Item "heads: 2192"
Although memory overhead of bigger buckets is kept inside the bucket, for
smaller buckets, it is kept in separate areas.  This field gives the
total size of these areas.
.ie n .IP """chain: 0""" 4
.el .IP "\f(CWchain: 0\fR" 4
.IX Item "chain: 0"
\&\fBmalloc()\fR may want to subdivide a bigger bucket into smaller buckets.
If only a part of the deceased bucket is left unsubdivided, the rest
is kept as an element of a linked list.  This field gives the total
size of these chunks.
.ie n .IP """tail: 6144""" 4
.el .IP "\f(CWtail: 6144\fR" 4
.IX Item "tail: 6144"
To minimize the number of \fBsbrk\fR\|(2)s, \fBmalloc()\fR asks for more memory.  This
field gives the size of the yet unused part, which is \fBsbrk\fR\|(2)ed, but
never touched.
.SH "SEE ALSO"
.IX Header "SEE ALSO"
perldebug,
perl5db.pl,
perlguts,
perlrun,
re,
and
Devel::DProf.