Re: core dumps

From: Graham Gilmore (gilmore@gmgate.vircom.com)
Date: 04/03/96


On Wed, 3 Apr 1996, JRhone wrote:

> 
> > personally, I got tired of messing around with cores, seeing as how 
> > all the time I spent trying to get a readable core was time that I 
> > wasn't fixing the damn problem.  So I just run a beta version in the 
> 
> true, but most of the really hidden problems and bugs dont show up unless
> yer up for a day or so and have a bunch of players on, and runnin the mud 
> under gdb gets annoying imho, getting a nice fat core on a crash would be
> prime so i could jump into gdb, show the stack, and go fix the prob rather
> than sitting in gdb waiting for a crash

	Personally I run gdb in the background from a script, I capture the 
output using nohup, and I pipe in a file containing a command list which tells
gdb what to do.  So I don't have to hang around watching gdb... if the MUD 
crashes, I just go take a look in nohup.out, and it contains the 'where' 
listing as well as the MUD logs right up to the point of the crash.  It also 
shows a few variables which may or may not be relevant in the context of 
the crash, but are sometimes useful, like ch->player.name etc.
	I didn't originally come up with this method, I'm something of a 
Linux dummy :)  Someone else came up with it for another MUD that had 
been running on the system, and I simply adapted the relevant bits to 
make a new script.

	Graham Gilmore



This archive was generated by hypermail 2b30 : 12/18/00 PST