Re: Patch for SocketPort/SocketService problem - was: Serious SocketService/SocketPort problems

[David Sugar]

That would be me, actually, and yes, I did...Originally, socketport and
socketservice were actually modeled on how the Bayonne channel driver
allocates threads and sessions to produce a poolable resource under a
shared service thread that is then invoked through event callback.  
SocketPort/SocketService simply represented a socket abstraction of this
device model in a different domain (sockets rather than devices).

On Mon, 25 Nov 2002, Prochnow, Christian wrote:

> Hi List,
> 
> regarding the SocketPort attach issue i've developed a patch.
> this patch adds the possibility to pass a SocketService NULL-ptr on 
> SocketPort 
> construction, so that the SocketPort-ctor does not attach itself.
> One can then call SocketPort::attach(SocketService*) in the inherited 
> SocketPort-constructor (or later) to attach the object to a SocketService.
> 
> It seems like the initial author of SocketPort/SocketService already had this 
> idea, cause the SocketPort::attach() function was declared, but not 
> implemented :)
> 
> Regards,
> Christian
> 
>