Synopsis: Automating GDB backtraces. October 96
From: erwin@pip.dknet.dk
To: MERC/Envy Mailing List <merc-l@webnexus.com>
Subject: Re: Segmentation Fault Problem

On Sat, 5 Oct 1996, Eric Kidder wrote:

> Well, maybe not the core file itself *shudder*, but a backtrace and
> info local from gdb would be nice :)

Here's how automatically send yourself a stack  backtrace using the startup 
file:

In your area directory, create a .gdbinit file containing:

bt
info locals


and in your startup file:

../src/envy bla bla

if ( -e core ) 
  gdb -batch ../src/envy core | mail your@address
  mv core core.`date +%s`
endif


gdb will read your binary file, your core file, then execute everything 
in .gdbinit. When finished, it exits, because you used the -batch switch.

The mv lines causes the core file to be moved to a file named 
core.timestamp, so that repeated crashes do not overwite cores. This 
might be dangerous though if you have limited disk space, should you 
accidentally use it all up

Note that under BSD-style Unices the core file is called something else 
(binaryfilename.core)


==============================================================================
Erwin Andreasen  Viby J, Denmark   Computer Science Student at Aarhus Business
erwin@pip.dknet.dk        <URL:http://pip.dknet.dk/~pip1773>           College
==============================================================================