On Sun, 2004-03-07 at 19:51, Stephen McConnell wrote:
> Scott Brickner wrote:
> > At the moment, 3.0. I'm flexible, though - I'll take anything that's
> > considered a "production level" release.
> 
> 
> First I would recommend rolling up to 3.2.5 (but that's optional).
> 
> Secondly - prior to some deep and meaningful thinking I would suggest 
> your use coordination.  What your describing is a situation that I've 
> had to deal with several times in the past - in effect you have a family 
> of objects and objects need an extra level of coordination.  Typically 
> this can be addressed by declaring a dependency from each member of the 
> family of components on a singleton coordinator instance.  The 
> coordinator is supplied to all of the members of the family, all of the 
> children register back with the coordinator during deployment, and when 
> the coordinator is happy that all of the children are 
> ready-willing-and-able, the coordinator trips a "go" event.  One thing 
> to be careful to do is to post each event in a separate thread (at least 
> that's what I've done in the past).
> 
> More generally speaking - the notion of families of components that are 
> tied one way or another is a issue that I'm come across on multiple 
> occurs and its something that is not expressed in the component model. 
> I think its a subject that should be considered my deeply - probably 
> deserves a post over on [EMAIL PROTECTED]
> 
> Anyway - do you think the addition of a coordination component will 
> resolve the problem your dealing with.  If no - please let us know and 
> we dig deeper.
> 

I think that'll probably work. It feels a little ugly, but hey - right
now I need a working server, not a pretty one.

Thanks for the suggestion.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to