On 2/16/07, Maarten Bosteels <[EMAIL PROTECTED]> wrote:
On 2/16/07, Greg Duffy <[EMAIL PROTECTED]> wrote:
>
> Rob,
>
> Could you elaborate on this a bit? I definitely agree that any selector
> pool
> implementation should be pluggable, especially because some people might
> not
> want pooling at all. I'd just like to know what functionality you want
> that
> is missing in Commons Pool so I can better evaluate our options. It's been
> a
> while since I've used Commons Pool, but I do remember finding some
> limitations.
>
> Does anybody else have recent experience with it?
commons-dbcp is database connection pool implemented on top of commons-pool.
I think it is widely used.
We are using it since more than 2 years have had no issues with it.
But I agree, that a pluggable pooling implementation would be a good idea.
commons pool is a pluggable pooling API
the default implementation is limited in the 1.x codebase but 2.0 is
much more flexible but the bottom line is that extensions are going to
needed for advanced functionality.
IMHO what's more important is that MINA makes it easy and flexible to
use alternative implementations. as with most of the commons APIs, the
key is using IoC to allow any implementation of the API to be used for
any pool. haven't looked too deeply at MINA's usage so i don't know if
code changes would be required.
<snip>
> On 2/16/07, Rob Butler <[EMAIL PROTECTED]> wrote:
> >
> > Trustin,
> >
> > Please make the pool implementation pluggable. Apache commons pool is
> > pretty basic and limited in functionality. Of course I never have time,
> but
> > at some point I hope to implement (or find) a much better pool
> > implementation.
it's hard to find any which are much better: many others are just less
well tested and less widely used. it's really easy to create a pool
that works well in theory.
IMHO it would make more sense to work to implement the required
features by extending the commons-pool API or working with the commons
team to improve that API.
- robert
