[Home] [Groups] - Message: [Prev in Group] [Next in Group]
2776: Re: [MUD-Dev] Persistance/stability
[Full Header] [Plain Text]
From: "Chris Gray" <cg@ami-cg.GraySage.Edmonton.AB.CA>
Newsgroups: nu.kanga.list.mud-dev
Date: Sat, 2 Aug 97 09:05:33 MST
Organization: Kanga.Nu
[Chris L:]
:[Miro S:]
:>MUSHes and MOOs save everything across the reboot. MOOs even save the
:>active processes (something Cold doesn't do, and I don't know about
:>MUSH).
:FWIW I save active processes at runtime (ie across crashes). Well,
:sorta -- there's some caching and delays on the writes. Upon restart
:the machine will resume exactly as of some state minutes before the
:crash.
Saving the state of everything without saving information about how that
state is being changed would kind of defeat the point of saving. I'm more
of an event-based model, but I don't actually save the list of pending
events. What I do is allow a machine to have an action which is called
on a server restart. Similarly there is one called on a shutdown. That
allows explicit setup to gracefully resume across shutdown/restart. For
crashes, I just go with an old, but consistent, copy of the whole db.
That can result in the timing of things changing a bit, but I doubt if
its really noticeable, since there are likely to be a lot of things
getting "undone" when restarting from a pre-crash snapshot.
--
Chris Gray cg@ami-cg.GraySage.Edmonton.AB.CA