[vworld-tech] Verb/action handling

[ceo]

Bruce Mitchener wrote:
> Brian Hook wrote:
> 
>> I'm trying to come up with a scheme that handles verbs uniformly that 
>> isn't a kludge or incredibly slow or error prone.
>>
> 
> This depends on details that you haven't provided at all. :)
> 
> You may also find it interesting to read up on the various bits of 
> research regarding management of areas of interest and all of the other 

Sections on this in Agora use the term "spatial" to describe some 
relevant approaches, IIRC (?).

It would also be helpful to have an idea of what rate you expect to see 
events generated at. For instance, a long time ago (read: I haven't 
looked at this for ages, and am now rusty :)) I implemented a roomless 
MUD based on subscribe/notify semantics for each major "sense" (not 
hard-coded; if you were concerned about performance, one approach was to 
increase the number of "senses" you were using, which would have the 
effect of reducing the number of subscribers to each).

Although I've seen references in recent years to experimental MUDs that 
have gone much further, FYI tHe basic scheme I used was that every event 
was delivered to everything interested in that "sense". The slighltly 
more advanced version :) used quadtrees and event-radiuses to 
efficiently reduce the number of potential subscribers that needed to be 
checked for each event-notification to a manageable number.

(...As an aside, I went from there into MMOG systems, where the rate of 
events is often so high you need to be considerably more cunning...)

For a low rate, and a design decision to do no "heavy" processing in the 
event-callbacks, there's no problem with doing a basic circle-based 
collision-detect on who/what needs to be informed. IME, for a "normal" 
MUD, there's no need to use any of the cunning techniques of RTS's etc 
(e.g. incremental notifications - retain data structures with the 
answers to "who can see THIS" pre-calculated and updated automatically 
whenever individual objects move).

Adam