SEQ crash, debug file?

[cheese_poker]

Hi. Sorry if this sounds newbish, but I am new to Linux.

Every once in a while, SEQ will "crash". No error messages. Nothing. It just stops running like I did the File/Exit. I can relaunch it just fine and it will work again.

My question :

Is there information I can look at to get a hint of what is causing the spontaneous combustion? It's nothing disasterous, but kind of annoying. If I can fix it, I'd like to. :)

Thanks in advance.

[Cryonic]

look for a file named core in the directory that you launched SEQ from. Then search the dev board for using a debugger (gdb I think) to extract the information from the core file and post it in the dev forum.

[belgarath01]

every time my SeQ crashes like that is because some one camped in zone or some other thing like that a corpse disappearing ect..

[EtherMage]

This kind of thing is the Linux equivalent of a GPF in Windows, minus the error messages. Here's what to do so you can get a decent picture of what happened to the devs so they can try and fix things:

1. Make sure you are running the latest ShowEQ. If you are even a few days old, a recent change might make your report useless.
2. Make sure you are set to produce 'core' files. To check this, type 'ulimit -c'. If that produces '0', put the line 'ulimit -c 10000' at the end of your /root/.profile file. If that file doesn't exist, just put that line there.
3. Have a policy of keeping a log of ShowEQ output. To do this, run ShowEQ as 'showeq 2>&1 | tee showeq.log'. This will pipe all output to the 'tee' command, which copies it both to the screen and to the log file. If you don't want the screen output, you can just use 'showeq >showeq.log 2>&1'.
3. If/when showeq bombs, it will create a file called 'core' or 'core.<numbers>'. This file records the state of ShowEQ's memory at the time of the crash. To make this file useful to the devels, you need to convert it to what's called a 'stack backtrace' which shows them what part of their code failed. To do this run a command like this: 'gdb -c <new core file> /path/to/installed/showeq'. If done right, this will produce lots of loading text, and then a gdb prompt. At that prompt, type 'bt'. Now, copy that text to a file. An easy way is to size your xterm so it all fits in one screen (if you increase the width, run the command again to rewrap), select it all, then open a file in an editor somewhere and middle-click. This will paste the information. Post that and the log of ShowEQ output here. Once you've posted the log+backtrace, you can delete the log/trace/core files.

BTW, I'm not a showEQ dev; however, I am a programmer, and trust me, a good stack backtrace/log combination is a faster than a thousand stories of how a crash happened at spotting the problem. So if you want to help make showeq more stable, you want to do these steps. It's probably the best thing an average user can contribute to the development of ShowEQ, aside from perhaps mapping.

-EtherMage