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: streaming outbound messages (was Re: [Vote] 1 Msg or 2)
Date Wed, 07 Feb 2001 02:19:47 GMT
I don't see how we can define this away.  It's perfectly valid for
an output handler to throw a fault - and if it's the very last thing
in the incoming message that causes it then we shouldn't have
sent back anything except a fault.
-Dug


Sam Ruby/Raleigh/IBM@IBMUS on 02/06/2001 08:54:08 PM

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

To:   axis-dev@xml.apache.org
cc:
Subject:  Re: streaming outbound messages (was Re: [Vote] 1 Msg or 2)



Doug Davis wrote:
>
> Just wondering... let's say we get everything streaming
> and let's say that its soooo good that the outgoing
> message actually starts to leave the last handler
> before the before all other handlers are done.  And
> let's say that one of the handlesr (not the last one
> in the chain) throws a fault.  At this point the
> transport listener (or sender depending on how you
> look at things) has already send back an rc of 200
> but it really needs to be a non-ok rc, doesn't it?
> How does this get handled?

Default fault.  Produces output, then changes it's mind?  Gets an
java.lang.OutOfMemoryError?

Since we are defining what the contract for a handler is, we can define
away the former.  Or more precisely, define such an occurance as a invalid
handler and treat it as an instance of the latter.  The latter can happen
without streaming.  In fact, arguably, it is more likely to do so.

- Sam Ruby




Mime
View raw message