Re: [Webwork-user] OT: AOP toolkit, was Why I like WebWork

[Rickard <ri...@dr...>]

Roger Holbrook wrote:
> and spent several days playing.  I just kept reading, and reading, and =
smiling ;)  It
> was hugely instructive - many, many thanks for this gift.
>=20
> So here we are, 6 months later, and you mention your AOP Toolkit -

What I have now is a rewrite of that stuff. It wasn't very scalable=20
(e.g. interceptor/aspect instances couldn't be reused even if they were=20
stateless), and making interceptors Proxy InvocationHandlers just wasn't=20
the best way to do it. What I have now is not only easy to work with,=20
it's scalable (stateless instances can be reused, and object caches can=20
be flushed, like EJB) and fast (during an invocation no objects are=20
created by the framework itself). Other than that the basic semantics is=20
the same as the old framework.

It'd probably be quite trivial to implement an EJB server using it. But,=20
then you'd miss all the good stuff that AOP gives you. As I said, being=20
able to just add interfaces/implementations to an object without=20
actually modifying any code is quite powerful. Ya can't do that in EJB,=20
fer shure.

>>The core of it is really generic, and entirely based on reflection/dyna=
mic proxies.
>=20
> Needless to say, I'm now just full of curiosity!
>=20
> Has the Toolkit progressed since the 'first go' ? And if so how?
> It was so simple in its initial conception, that almost any change woul=
d imply a
> conceptual adjustment? etc. etc. ??????????????
>=20
> I for one, would love to see the current version being opensourced!

See above, it's a lot better/simpler. Who would've thought that possible..

Anyway, we've talked about doing the AOP toolkit as a separate thing=20
from our main product (which is a content management system). We'd like=20
to test it in our own stuff first, to see if it can be used in a large=20
system effectively. We think so, but need to try to actually know so :-)

/Rickard

--=20
Rickard =D6berg