axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Glen Daniels <>
Subject RE: Removing boxcarring code
Date Thu, 09 May 2002 17:33:18 GMT

I just don't like the boxcarring thing in general, since it's underspecified and there seem
to be all kinds of potential problems - if you have three methods in the same body, two of
which succeed followed by a fault, what do you get back?

Although something like what you propose could work for this particular case, there may be
other cases in a boxcar-enabled world where people have body elements that they intend to
be executed that are referenced by some header, and therefore would have been touched by deserialization....
 I'd still prefer to remove the code from RPCProvider, and allow anyone who wants to build
a "BoxcarProvider" which does this kind of thing.

As for the "real root" issue, I agree that's an issue, but the colloquial solution which everyone
uses is that the first element is the "real" one in cases where you can't tell via root.


> -----Original Message-----
> From: Doug Davis []
> Sent: Thursday, May 09, 2002 12:02 PM
> To:
> Subject: Re: Removing boxcarring code
> I should also point out that I think there is a fix for this...
> As we work our way thru the root body elements only process 
> the ones that
> have not been touched by any previous deserialization 
> processing.  This
> should fix this bug too.
> -Dug
> Two things about this:
> 1 - While I don't know for sure people are using it I have had several
> people ask me about it so it might be in use.
> 2 - If the serialization happened slightly differently and the complex
> object appeared before the RPC call then w/o the root 
> attribute how can
> anyone know which root is the real root?  Seems like we're 
> just ignoring
> future problems with a short-term fix.
> That being said -0.
> :-)
> -Dug
> Glen Daniels <> on 05/09/2002 11:43:54 AM
> Please respond to
> To:    "'Axis-Dev (E-mail)'" <>
> cc:
> Subject:    Removing boxcarring code
> Please take a look at this bug.  I'm proposing we remove the loop in
> RPCProvider which attempts to execute each element in the 
> soap body.  Does
> anyone have any problems with this?
> If not, I'll plan to go ahead with this by the end of the day 
> today, with
> the potential to roll the change back if there's a great 
> outcry after that.
> --Glen

View raw message