OK, I ma going to try to get some time in tomorrow. I pretty much HAVE to do some things for Dave tomorrow, so I will be infront of my computer not working on work work, so I can do some pokin around and thinking myself



On Nov 4, 2006, at 6:59 PM, Barney Boisvert wrote:

A bit of digging seems to further point at ColdSpring.  I've attached
(though don't know if it'll go through) an example mimicking the
non-advice proxy path through AutoProxyBean.callMethod, and the
internally invoked calls to the target instance do not go through the
proxy, only externally invoked calls do.  The implementation is
totally non-general, but the structure is the same, near  as I can
tell.  It should be zero-dependancy, just unzip and run.

I haven't looked into the advice chain path yet.  The concept is
clear, but I haven't grokked the implementation yet, so I can't speak
for that side of things (though I still suspect that's the culprit).
I'll probably dive into that at some point, but not this second.

cheers,
barneyb

On 10/27/06, Chris Scott <[EMAIL PROTECTED]> wrote:

Interesting. I think this is not really the result of how ColdSpring creates aop proxies (which are proxies of the beans, btw) it is really perhaps the result of how cf works itself. In reality since there is no such thing as a cfc 'object', and all the methods you define for a cfc become their own command objects anyway, it mayt seem that there there really only ever is a proxy of a collection of methods, as opposed to a true proxy of an object. Maybe I can try to introduce a concept of a 'caller' object to try to alleviate this


Chris



-- 
Barney Boisvert
360.319.6145

Got Gmail? I have 100 invites.
<proxies.zip>

Reply via email to