axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Doug Davis" <...@us.ibm.com>
Subject Re: transport chains
Date Sat, 28 Jul 2001 20:49:01 GMT
I agree that we'll return to the transport listener but I seem to remember
there was a case where we wanted to actually change the handlers invoked on
the response chain based on the web service itself.  It's not a huge thing
(or even a very likely case) I just noticed it in the code.
-Dug


"Glen Daniels" <gdaniels@macromedia.com> on 07/28/2001 04:44:07 PM

Please respond to axis-dev@xml.apache.org

To:   <axis-dev@xml.apache.org>
cc:
Subject:  Re: transport chains



IIRC, at the F2F we decided that you do always want to go back through the
transport-specific response chain, and that if you want to send a response
out a different transport sender, you do that yourself and then leave the
response message in the MessageContext empty.  This allows req/resp
protocols such as HTTP to ensure that the right thing (i.e whatever is
configured) always happens with a response, even if that response is empty.

I'm not particularly averse to either way of doing it (except I don't
really
like unnecessary registry lookups such as we had before) - I'd say we leave
it as is for now and then come up with a list of use-cases that demonstrate
the breadth of what people might want to do with this stuff.  That might
shed some more light on whether one approach or the other is "better".

--G

----- Original Message -----
From: "Doug Davis" <dug@us.ibm.com>
To: <axis-dev@xml.apache.org>
Sent: Saturday, July 28, 2001 4:29 PM
Subject: Re: transport chains


> If on the fly (or based on the specific web service) a different response
> transport chain is desired - maybe a handler in the web service specific
> wants to remove the entire transport specific response chain because
based
> on the specific web service invoked it no longer needs it - ie. that web
> service had it's own transport sender to send back the response a
different
> way and it doesn't want the normal transport specific response chain
> invoked since there won't be any message to send back.  I don't think
that
> was the exact scenario we talked about at the f2f - but I think it was
> close and I'm sure there are other cases.
> -Dug
>
>
> Rob Jellinghaus <robj@unrealities.com> on 07/28/2001 04:22:42 PM
>
> Please respond to axis-dev@xml.apache.org
>
> To:   axis-dev@xml.apache.org, axis-dev@xml.apache.org
> cc:
> Subject:  Re: transport chains
>
>
>
> What's the use case here?  Seems a bit suspicious to me....
>
> Cheers,
> Rob
>
> At 02:58 PM 7/28/2001 -0400, Doug Davis wrote:
> >right now we choose the transport chain at the start of processing in
the
> >server, and then we call the request side of it.  Later on we will call
> the
> >response side - at one point we talked about allowing a handler to
change
> >the outgoing transport response chain by setting a field in the msg
> >context.  Currently in the code you can't do this because the transport
> >chain isn't "re-found" before the response side is called.  Do we want
to
> >change this so we "re-find" it - allowing the response side of the
> >transport chain to be changed on the fly?
> >-Dug
> >
> >
> >
>
>
>




Mime
View raw message