On 3/5/06, Chris Scott <[EMAIL PROTECTED]> wrote:
> Nice work Sean! Funny we had talked about the issues of traversing
> through inheritance trees a while back, and I think there were some
> philosophical discussions of wether or not ColdSpring and AOP SHOULD
> do it. We had decided against it, but 6 months ago we weren't playing
> with reactor. Along the same lines, the error I purposefully threw
> about not declaring a return type is probably going to be trumped by
> a changing coldfusion landscape.

Yeah, it's interesting how things have changed.

> I was actually thinking of making
> this change after we chatter earlier today, but I'm busy preparing
> for cfObjective, so it was going to have to wait. Thanks for taking
> the load off! Are you going to commit it to cvs?

I wanted to run it up the flagpole here first to see if anyone had
grievous objections. I'll probably commit the return type change and
the autowire / setterType changes but I need to look a little deeper
into the metadata walking stuff. Right now I have everything I need
working with just that one change to createProxyInstance() but I'm
concerned that I'm missing something in other use cases (that
currently walk metadata.functions).

The other roadblock I've hit is also somewhat philosophical: I really
want to apply AOP to Reactor record objects - but those are stateful
and are created on the fly during each request. AOP feels too
heavyweight for that scenario. I know ColdSpring isn't really designed
for managing transient stateful objects like that...
--
Sean A Corfield -- http://corfield.org/
Got frameworks?

"If you're not annoying somebody, you're not really alive."
-- Margaret Atwood

Reply via email to