December 1999
- Neural Networks as the AI system for a MUD? PartyG2816@aol.com
- Neural Networks as the AI system for a MUD? claw@kanga.nu
- Neural Networks as the AI system for a MUD? Marc Bowden
- Neural Networks as the AI system for a MUD? Lo, Kam
- Neural Networks as the AI system for a MUD? Marc Hernandez
- Neural Networks as the AI system for a MUD? Kræn Munck
- Neural Networks as the AI system for a MUD? Hans-Henrik Staerfeldt
- Neural Networks as the AI system for a MUD? PartyG2816@aol.com
- Neural Networks as the AI system for a MUD? Hans-Henrik Staerfeldt
- Neural Networks as the AI system for a MUD? Andrew Norman
- Neural Networks as the AI system for a MUD? Mik Clarke
- New member IronWolf
- Scenarios Chris Lloyd
- ColdStore J C Lawrence
- New AI Engine released Fabian
- New AI Engine released Steve Houchard
- New AI Engine released Bruce Mitchener, Jr.
- New AI Engine released Elysium
- AI's in MUDS and Online Gaming IronWolf
- AI's in MUDS and Online Gaming Matthew Mihaly
- AI's in MUDS and Online Gaming IronWolf
- AI's in MUDS and Online Gaming Ryan P.
- Garbage Collection Miroslav Silovic
Cynbe ru Taren <cynbe@muq.org> writes:
> In Muq, I'm currently using a plain-jane mark-and-sweep monolithic
> gc I hacked together one weekend just to have something rather than
> nothing.
Hmm. :) Problem with mark&sweep is bad realtime performance. You can
look to 1-2 seconds pause per 10MB allocated. Not too bad unless you
have *lots* of active objects or are running realtime combat. See
BattleTech MUSHes for realtime combat example (and the ammount of
annoyance 1 second lag can cause).
> I've been contemplating going to a two-generation system and/or
> implementing Dijkstra's three-color incremental algorithm.
Generational GC tends to require the same infrastructure as tricolor
(i.e. write barrier).
> Is there prior mud art with other approaches? Are there working
> systems or good references I should be boning up on before proceeding?
No MUD systems that I know of, but there are plenty of programming
languages implementations that use good GC systems. Note that GC has
one serious drawback: it tends to touch all the pages it scans. While
it's possible to check whether the page is on swap (and then try and
be clever about scanning the pointers from it), GCs in general don't
do this, and so GC systems don't perform well when the machine begins
to thrash.
My favorite GC implementations are Hans Boehm's (major problem: uses
VM tricks to handle generationality), rscheme's (check
www.rscheme.org) and the one in TOM (incremental, nongenerational for
now, and with very clean interface to the object system that allows
objects a good degree of control over the way they get collected).
So here are the MUD/GC issues:
- Using well-implemented GC results in a code that doesn't have to
worry at all about memory bugs - you probably want conservative stack
scanning if you're doing *any* embedded C, but you can almost always
scan heap exactly. Also, you can use circular datastructures without
limits, work with graphs in a general manner (refcounting tends to
break with general graphs) and keep all the memory issues separate
from your object system (i.e. no worries where/when
constructors/destructors get called). You also have symmetry between
objects pointed from stack and the objects pointed from heap - this is
VERY helpful for design. Note that GC may treat stack and heap
pointers differently, the symmetry is in the GC interface.
- Write barrier... This is problematic part. In C, calling write
barrier for heap objects means you have to do
GC_ASSIGN_POINTER(object->foo, another_object);
or
object->foo = another_object;
GC_LINK(object, another object);
(you need to know inter-generational pointers in gengc, that's what
the second parameter is for).
rather than just
object->foo = another_object;
This is easy to forget. C++ with some templates has trivial workaround
(just implement write barrier into assignment operator) but will tend
to call write barrier WAY too often. Alternatively one could probably
program a lint-like utility to statically check for this problem.
- swap interaction. This topic, to put it plainly, sucks. One way to
treat this problem is to treat parts of your object tree as separate
systems and collect them using distributed GC technique - it handles
one hard problem by turning it into another hard (but more researched)
problem.
- one-bit refcounting: Some datastructures are VERY transient in
nature - strings, for instance, or closures when you use them just to
avoid enumerators. To handle them, you give them one bit refcount: You
know when these will disappear from stack, so you can use a single-bit
flag that checks whether this object is referenced from heap. If no,
once it disappears from the local scope, you can kill it. This doesn't
suffer from threading interactions typical for refcounting, and can
drastically increase performance if you allocate your strings on a
GC'ed heap.
- distributed gc: This becomes relevant if you're running your mud on
several servers. I haven't thought much about it, apart from noting
that TOM supports somewhat limited form of it (read: VERY
fault-intolerant).
More information on this topic: ftp://ftp.cs.utexas.edu/pub/garbage/
--
How to eff the ineffable? - Capture of players. (was: Fair/Unfair? Scenarios) Kræn Munck
- Capture of players. (was: Fair/Unfair? Scenarios) J C Lawrence
- Capture of players. (was: Fair/Unfair? Scenarios) Chris Lloyd
- Biosystems (was Fair/Unfair? Scenarios) Richard Ross
- Biosystems (was Fair/Unfair? Scenarios) Philip Loguinov -- Draymoor
- Biosystems (was Fair/Unfair? Scenarios) Mik Clarke
- Biosystems (was Fair/Unfair? Scenarios) J C Lawrence
- Biosystems (was Fair/Unfair? Scenarios) Greg Miller
- Biosystems (was Fair/Unfair? Scenarios) Dundee
- Biosystems (was Fair/Unfair? Scenarios) Travis S. Casey
- Biosystems (was Fair/Unfair? Scenarios) Douglas Couch
- Biosystems (was Fair/Unfair? Scenarios) Marc Hernandez
- Biosystems (was Fair/Unfair? Scenarios) J C Lawrence
- Biosystems (was Fair/Unfair? Scenarios) Mik Clarke
- Ideas for dynamic worlds Nolan Darilek
- Ideas for dynamic worlds Chimera
- Ideas for dynamic worlds Ilya, Game Commandos
- Ideas for dynamic worlds J C Lawrence
- Ideas for dynamic worlds Nolan Darilek
- Ideas for dynamic worlds Joe Kingry
- Ideas for dynamic worlds David Morgan
- The GTS Library J C Lawrence
- Combat systems Neil Edwards
- Combat systems Kylotan
- Combat systems Kylotan
- Combat systems Chris Lloyd
- Combat systems J C Lawrence
- Combat systems Marc Spoorendonk
- Combat Systems Ben
- Combat Systems Raph Koster
- Metakit J C Lawrence
- Commercial Online Roleplaying Games (fwd) J C Lawrence
- Mud-Dev FAQ Marian Griffith
- Mass bannings (was The grass is always greener in the other field) AR Schleicher
- Game Balance: Statistical Analysis in MPORPGs Lovecraft
- Game Balance: Statistical Analysis in MPORPGs Koster, Raph
- Biosystems & simulation [RAMBLE] Ben Greear
- Souveniers Douglas Couch
- Souveniers Lovecraft
- Souveniers J C Lawrence
- Souveniers Raph & Kristen Koster
- Souveniers J C Lawrence
- Souveniers Raph & Kristen Koster
- Ideas for dynamically generated worlds Cynbe ru Taren
- Ideas for dynamically generated worlds J C Lawrence
- Balancing between anxiety and boredom (was Fair/Unf air? Scenarios (fwd) ) Sellers, Michael
- Online Migration and population mobility in a virtual gaming setting - Ultima Online J C Lawrence
- lockpicking Matthew Mihaly
- Optimized Object Storage Ian Macintosh
- Optimized Object Storage Sellers, Michael
- Optimized Object Storage Matthew Mihaly
- Optimized Object Storage J C Lawrence
- Optimized Object Storage Ian Macintosh
- Optimized Object Storage Ian Macintosh
- Optimized Object Storage Sellers, Michael
- Optimized Object Storage Ian Macintosh
- The Habitat Papers are missing J C Lawrence
- Waving Hands -- Debian's Spellcast for Linux J C Lawrence
- Waving Hands -- Debian's Spellcast for Linux Dan Shiovitz
- Waving Hands -- Debian's Spellcast for Linux Travis Casey
- Waving Hands -- Debian's Spellcast for Linux J C Lawrence
- Upload to ftp.kanga.nu J C Lawrence
- Farewell again Ilya, GC
- Telnet Protocol and Winsocks method
- Telnet Protocol and Winsocks J C Lawrence
- Telnet Protocol and Winsocks cg@ami-cg.GraySage.Edmonton.AB.CA
- Telnet Protocol and Winsocks Ilya, GC
- Telnet Protocol and Winsocks J C Lawrence
- Proper liscense for MUD source? Perhaps not GPL... (fwd) J C Lawrence
- Proper liscense for MUD source? Perhaps not GPL... (fwd) J C Lawrence
- Proper liscense for MUD source? Perhaps not GPL... (fwd) Ben Greear
- Proper liscense for MUD source? Perhaps not GPL... (fwd) Christopher Allen
- Proper liscense for MUD source? Perhaps not GPL... (fwd) J C Lawrence
- Proper liscense for MUD source? Perhaps not GPL... (fwd) Christopher Allen
- MUD source licensing: beyond GPL? (fwd) J C Lawrence
- MUD source licensing: beyond GPL? (fwd) Matthew Mihaly
- Classes and Races and more (a BIG list) (fwd) J C Lawrence
- Originality/Points of Reference (was Classes and Races and more (a BIG list) (fwd)) Richard Ross
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Justin Rogers
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) Justin Rogers
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Christopher Kohnert
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Matthew Mihaly
- Collecting ideas for a MUD server... (fwd) Jon A. Lambert
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) Wesley W. Terpstra
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) Wesley W. Terpstra
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) Greg Miller
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Collecting ideas for a MUD server... (fwd) Rahul Sinha
- Collecting ideas for a MUD server... (fwd) J C Lawrence
- Originality/Points of Reference Ian Klimon, Esq.
- Job Openings - Mud Engineering Christopher Allen
- PGP player certificates (was: collecting ideas...) Wesley W. Terpstra
- PGP player certificates (was: collecting ideas...) David Bennett
- Re[4]: The grass is always greener in the other field Travis Casey
- Re[4]: The grass is always greener in the other field J C Lawrence
- Re[4]: The grass is always greener in the other field Adam Wiggins
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) Wesley W. Terpstra
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) J C Lawrence
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) Marc Bowden
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) Greg Miller
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) cg@ami-cg.GraySage.Edmonton.AB.CA
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) Hans-Henrik Staerfeldt
- Two threads forced to one CPU? (was: Collecting ideas for a MUD server...) cg@ami-cg.GraySage.Edmonton.AB.CA
- PGP confusions hopefully resolved (was: collecting ideas ...) Wesley W. Terpstra
- MUD-Dev digest, Vol 1 #237 - 9 msgs Dr. Cat
- MUD-Dev digest, Vol 1 #237 - 9 msgs J C Lawrence
- MUD relevant references (was: The grass is always greener...) Ola Fosheim Grøstad
- Re[6]: The grass is always greener in the other field Travis Casey
- Embedded languages, object persistance... ack. Joe Kingry
- Embedded languages, object persistance... ack. cg@ami-cg.GraySage.Edmonton.AB.CA
- Embedded languages, object persistance... ack. Laurent Bossavit
- Embedded languages, object persistance... ack. Kevin Littlejohn
- Embedded languages, object persistance... ack. J C Lawrence
- Embedded languages, object persistance... ack. J C Lawrence
- Embedded languages, object persistance... ack. Jay Carlson
- Embedded languages, object persistance... ack. Jay Carlson
- Embedded languages, object persistance... ack. J C Lawrence
- Embedded languages, object persistance... ack. icecube@ihug.co.nz
- Embedded languages, object persistance... ack. Kevin Littlejohn
- Storing tokens with flex & bison Christer Enfors
- Storing tokens with flex & bison Rahul Sinha
- Storing tokens with flex & bison J C Lawrence
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Christer Enfors
- Storing tokens with flex & bison Ola Fosheim Grøstad
- Storing tokens with flex & bison J C Lawrence
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Kevin Littlejohn
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Phillip Lenhardt
- Storing tokens with flex & bison Dominic J. Eidson
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Chris Jones
- Storing tokens with flex & bison J C Lawrence
- Storing tokens with flex & bison cg@ami-cg.GraySage.Edmonton.AB.CA
- Storing tokens with flex & bison Ola Fosheim Grøstad
- Storing tokens with flex & bison Per Vognsen
- Storing tokens with flex & bison Christer Enfors
- Storing tokens with flex & bison Chris Turner
- Storing tokens with flex & bison Christer Enfors
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison Christer Enfors
- Storing tokens with flex & bison Jon A. Lambert
- Storing tokens with flex & bison J C Lawrence
- Storing tokens with flex & bison Greg Miller
- Storing tokens with flex & bison J C Lawrence