Leo Sutic wrote on Friday, October 24, 2003 10:03 AM:
> What you describe would be per-thread provided that you have
> one thread = one request.

Nope. That was not enough. To use the component again, we would have to reinitzialize 
it (unless we were in the same scope and would make use of caching). Since the thread 
lives longer than the request, the objects returned by the pool were in indefinate 
state. OTOH I suppose that this was the basic problem, we should not have components 
with state at all, but I switched the company and therefore I don't care anymore 
anyway <g>

> It also makes me think about what
> expectations clients have in regards to lookups. More and
> more, it seems like the lifestyle isn't a property of the
> component impl, but a combination of impl and client expectations.
> 
> Is this too strong coupling, or legitimate use?

This is the normal problem with generalization and possibly wrong usage of the 
concepts :)

Regards,
J�rg

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

Reply via email to