Re: ROLE & Lifecycles in C#

[Stephen McConnell]

Geer, Christopher S wrote:

 

-----Original Message-----
From: Berin Loritsch [mailto:[EMAIL PROTECTED]
Sent: Wednesday, July 30, 2003 9:14 AM
To: Avalon Developers List
Subject: Re: ROLE & Lifecycles in C#
Geer, Christopher S wrote:
   

If there is already a IDisposable.Dispose(), then why not use it?
       

What

     

context is it used with?  I see no sense in reinventing the wheel.
       

I agree we should use the provided interface if we can.

     

Short list is:

Configurable
LookupEnabled
Initializable
Disposable
       

I'm assuming that you mean to role start() and stop() into
     

Initializable
 

or something? Also, what about Suspendable, Reconfiguration and the
other Re's? Personally I beleive that those are very important
interfaces to support. While you could just role those into the list
above that would almost require everyone support them, which would
     

mean
 

if someone didn't want them they would just have empty methods,
     

giving
 

someone a false belief that the compoenent actually supported them.
     

All the Re's should be removed until we decide what to do with them.
Noone has come up with a good idea for any of the Re's or how to
   

support
 

them.  They are legacy interfaces that in hindsight would have been
   

better
 

to nix back when we made the Avalon 4.0 baseline.
   

Getting rid of these for now doesn't really bother me too much.

 

In the future, when we have a requirement for something along these
   

lines
 

we can discuss how to do it, what the contracts should be, and then
   

expand
 

the framework.

As to Startable [start() and stop()], it seems redundant when we have
Initializable and Disposable.  Again, Suspendable is one of those
   

things
 

that loses any kind of usefulness if there is no reason to temporarily
suspend a component (reconfiguration, etc.).
   

I disagree. I think Startable is an absolutely necessary interface.

I agree.

There are perfectly valid cases where components may not need to be
"initialized" but do need to be started and stopped. Also, stop() and
dispose() have two different functions also, in my opinion. For example
(and this might not be the best example but...) if you were building a
web server, you might want to stop the web server from taking requests
and then collect metrics from the server (while its not changing) before
disposing of it.
The start stop pair make total sence when you component is a server 
style component.  Initialization is the point where the component 
validates itself relative to everything supplied to it - and start 
signals to the component to commence the process of handling doing 
whatever it does - normally handling requests or whatever.

While you could wrap the start() method into the Initializable interface
if we reuse the .NET IDisposable interface we won't be able to wrap the
stop() method into it.
 

What's the problem with keeping the notions seperate?

<snip>

I think there needs to be a way to modify a components configuration without disposing of it and loosing its state information.

Agree.

Cheers, Steve.

--

Stephen J. McConnell
mailto:[EMAIL PROTECTED]
http://www.osm.net
Sent via James running under Merlin as an NT service.
http://avalon.apache.org/sandbox/merlin


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