Today, Lord of Chaos spake to the list:
> Probably.. :) Now to figure out what it is, I'm usually pretty
> carefull about not creating non-endable loops.
Well, if you ever catch the problem in action, get its process id (PID)
from 'ps' and do, "kill -s 11 <pid>" filling in for <pid> as appropriate.
This will cause your MUD to crash, reporting that it experienced a
segmentation fault and do a core dump. Now you can see exactly where you
were when the problem was occuring. If you are having difficulty catching
the problem, you can probably write a small program to periodically test
the MUD and automatically perform the kill. Since we aren't quite sure
exactly what your problem is yet, YMMV.
If all this seems too much work, begin inserting logging statements into
game_loop and various activity functions. Check the additions you've made
since you can first remember the problem. Some near-infinite loops are
difficult to see at first glance. The idea is not to find it exactly with
the logging statements, but your choices down so you can more closely
inspect individual blocks of code.
If this last step seems like too much work, you're not using a revision
control system (RCS, CVS, etc.) and you should begin to do so now. The
ability to back out, review, and keep extensive information on your
changes is a vital part of any ongoing project. The failure to use these
tools is ultimately the reason for so many muds failing. While not every
pocket of amateur software developers need such tools, mud programmers
need them just as much, if not more than, most professional software
developers. Erwin S. Adreasen, I believe, wrote a treatise on the use of
RCS for mud programmers that should be available on Ceramic Mouse. As
you're undoubtedly already aware, we use CVS for official development of
CircleMUD. Learn it, love it, live it.
Good luck.
-dak
+------------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://qsilver.queensu.ca/~fletchra/Circle/list-faq.html |
+------------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/15/00 PST