Re: checkpointing

From: Gary Barnett (gbarnett@POLARNET.COM)
Date: 04/06/98


-----Original Message-----
From: Angus Mezick <angus@EDGIL.CCMAIL.COMPUSERVE.COM>
To: CIRCLE@post.queensu.ca <CIRCLE@post.queensu.ca>
Date: Monday, April 06, 1998 5:03 AM
Subject: Re:  checkpointing


>um... checkpointing is mostly for catching infinite loops.  On a heavy
use/slow
>machine, saving your whole world will take a bit of time.  it is set to 3
>minutes so that the problem HAS to be an infinite loop, and not just some
god
>command that is a bit intensive.
>--Angus


C'Mon.. Didn't you read the message?

Find me a routine that takes longer than 30 seconds to run.. Bottom line is
that 30 seconds is way out of the range for any set of routines in a
realtime program.

Saving the whole world? On a heavily used machine? That's why I suggested
the _default_ be lowered and anyone who comes up with a heavily used machine
or a routine written poorly enough to not return control in 15 seconds would
have to change _their_ checkpoint timeout.

I just checked my machine. Saving the entire world (all mobs/all objs/all
rooms) in a loop (which I had to construct, as I never save the entire world
at one time *boggle*) .. took less than 3 seconds. Admittedly I have a fast
machine...

When was the last time the checkpointing value was updated to compensate for
new faster hardware in common use? I know it hasn't been since bpl11.. or
even 2.x if memory serves.

I changed mine to 15 seconds.. haven't had any problems.. and I recommend
others do the same, especially if they are running on the same machine I am.
:-)

--Mallory


     +------------------------------------------------------------+
     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     | http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
     +------------------------------------------------------------+



This archive was generated by hypermail 2b30 : 12/15/00 PST