NAME
bashdb - bash debugger script
SYNOPSIS
bashdb [options] [--] script-name [script options]
bashdb [options] -c execution-string
bash --debugger [bash-options...] script-name [script options]
DESCRIPTION
"bashdb" is a bash script to which arranges for another bash script to
be debugged. The debugger has a similar command interface as gdb(1).
The way this script arranges debugging to occur is by including (or
actually "source"-ing) some debug-support code and then sourcing the
given script or command string.
One problem with sourcing a debugged script is that the program name
stored in $0 will be "bashdb" rather than the name of the script to be
debugged. The debugged script will appear in a call stack not as the
top item but as the item below "bashdb". If this is of concern, use the
last form given above, "bash --debugger" script-name [script-options].
If you used bashdb script and need to pass options to the script to be
debugged, add "--" before the script name. That will tell bashdb not to
try to process any further options.
See the reference manual <http://bashdb.sourceforge.net/bashdb.html>
for how to to call the debugger from inside your program or arrange for
the debugger to get called when your program is sent a signal.
OPTIONS
-h | --help
Print a usage message on standard error and exit with a return code
of 100.
-A | --annotation level
Sets to output additional stack and status information which allows
front-ends such as emacs to track what’s going on without polling.
This is needed in for regression testing. Using this option is
equivalent to issuing:
set annotation LEVEL
inside the debugger.
-B | --basename
In places where a filename appears in debugger output give just the
basename only. This is needed in for regression testing. Using this
option is equivalent to issuing:
set basename on
inside the debugger.
-n | nx
Normally the debugger will read debugger commands in
"~/.bashdbinit" if that file exists before accepting user
interaction. ".bashdbinit" is analogus to Perl’s ".perldb" or GNU
gdb’s ".gdbinit": a user might want to create such a debugger
profile to add various user-specific customizations.
Using the "-n" option this initialization file will not be read.
This is useful in regression testing or in tracking down a problem
with one’s ".bashdbinit" profile.
-c command-string
Instead of specifying the name of a script file, one can give an
execution string that is to be debugged. Use this option to do
that.
If you invoke the debugger via "bash --debugger", the filename that
will appear in source listing or in a call stack trace will be the
artifical name *BOGUS*.
-q | --quiet
Do not print introductory version and copyright information. This
is again useful in regression testing where we don’t want to
include a changeable copyright date in the regression-test
matching.
-x debugger-cmdfile
Run the debugger commands debugger-cmdfile before accepting user
input. These commands are read however after any ".bashdbinit"
commands. Again this is useful running regression-testing debug
scripts.
-L | --library debugger-library
The debugger needs to source or include a number of functions and
these reside in a library. If this option is not given the default
location of library is relative to the installed bashdb script:
"../lib/bashdb".
-T | --tempdir temporary-file-directory
The debugger needs to make use of some temporary filesystem storage
to save persistent information across a subshell return or in order
to evaluate an expression. The default directory is "/tmp" but you
can use this option to set the directory where debugger temporary
files will be created.
-t | --tty tty-name
Debugger output usually goes to a terminal rather than stdout or
stdin which the debugged program may use. Determination of the tty
or pseudo-tty is normally done automatically. However if you want
to control where the debugger output goes, use this option.
-V | --version
Show version number and no-warranty and exit with return code 1.
-X | --trace
Similar to ""set -x"" line tracing except that by default the
location of each line, the bash level, and subshell level are
printed. You might be able to get something roughly similar if you
set "PS4" as follows
export PS4='(${BASH_SOURCE}:${LINENO}): ${FUNCNAME[0]}\n'
In contrast however to ""set -x"" tracing, indentation of the
original program is also preserved in the source output. And if you
interrupt the program with a break (a "SIGINT" signal), you will go
into the debugger (assuming your program doesn’t trap "SIGINT").
BUGS
The "bashdb" script and "--debugger" option assume a version of bash
with debugging support. That is you can’t debug bash scripts using the
standard-issue version 2.05b bash or earlier versions. In versions
after 3.0, debugging should have been enabled when bash was built. (I
think this is usually the case though.) If you try to run the bashdb
script on such as shell, may get the message:
Sorry, you need to use a debugger-enabled version of bash.
Debugging startup time can be slow especially on large bash scripts.
Scripts created by GNU autoconf are at thousands of lines line and it
is not uncommon for them to be tens of thousands of lines.
There is a provision to address this problem by including a fast file-
to-array read routine (readarray), but the bashdb package has to be
compiled in a special way which needs access to the bash source code
and objects.
Another reason of the debugger slowness is that the debugger has to
intercept every line and check to see if some action is to be taken for
this and this is all in bash code. A better and faster architecture
would be for the debugger to register a list of conditions or stopping
places inside the bash code itself and have it arrange to call the
debugger only when a condition requiring the debugger arises. Checks
would be faster as this would be done in C code and access to internal
structures would make this more efficient.
SEE ALSO
· <http://bashdb.sourceforge.net/bashdb.html> - an extensive
reference manual.
· <http://bashdb.sourceforge.net> - the homepage for the project
· <http://www.gnu.org/software/bash/manual/bashref.html> - bash
reference manual
AUTHOR
The current version is maintained (or not) by Rocky Bernstein.
COPYRIGHT
Copyright (C) 2003, 2006, 2007 Rocky Bernstein
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
$Id: bashdb-man.pod,v 1.10 2009/06/22 22:41:10 rockyb Exp $