Re: [Asyncweb client] a number of changes to migrate from the Geronimo sandbox

["이희승 (Trustin Lee)" <tr...@gmail.com>]

2008-03-04 (화), 13:50 -0700, Mike Heath 쓰시길:
> Alan D. Cabrera wrote:
> > 
> > On Mar 3, 2008, at 12:26 PM, Sangjin Lee wrote:
> > 
> >> We have been making a number of changes on the Geronimo sandbox side
> >> for the
> >> AsyncHttpClient.  We've been meaning to move them over, but the pent-up
> >> changes have grown to quite a few.  I haven't had time to sit down and
> >> analyze whether certain bug fixes are applicable for the asyncweb
> >> codec.  I
> >> hope I'll be able to do that some time this week or next week at the
> >> latest.
> >> Just a heads up for you...
> > 
> > Has the merge for Geronimo AHC been completed on the Mina side yet? 
> > IIRC, someone was still integrating the AsyncWeb codec.
> 
> No it's still in limbo.  I haven't had as much time as I anticipated to
> work on this.  On top of the fact that is taking much more time than I
> originally estimated that it would (doesn't it always though?)
> 
> The biggest problem is that AHC uses the HttpRequest object to store the
> state of the request, the  future, the callback, etc.  This doesn't work
> at all with the AsyncWeb Commons code.  Unfortunately, this also affects
> almost everything in AHC so it's not something that can be easily
> changed in small steps.
> 
> In my local working copy, I've moved everything over to use the AsyncWeb
> commons HttpRequest and I've moved all the state information into the
> AHC ResponseFuture object.  This has, of course, required API changes.
> Everything compiles, but nothing works and it will take a lot more
> effort to make it work.
> 
> I'm not sure how best to proceed.  We need to port support for HTTP
> proxy and HTTP auth to AsyncWeb Commons.  There are also a lot of AHC
> tests that I think would be valuable to have in AsyncWeb Commons.
> 
> After that, honestly, I think that we should focus our efforts on the
> new client API's that are being discussed.  It doesn't make sense to
> port AHC over to use AsyncWeb Commons if were only going to go and
> revamp the API anyway.
> 
> We can still capitalize on a lot of the AHC work in AsyncWeb Commons and
> this can be done incrementally.
> 
> Thoughts?

I agree with you that HttpRequest and HttpResponse should remain as
stateless form.  HttpResponseFuture also sounds good to me.  Also, it's
a good idea to make things to work step by step.
-- 
Trustin Lee - Principal Software Engineer, JBoss, Red Hat
--
what we call human nature is actually human habit
--
http://gleamynode.net/