Hi,
> > It seems to me that HessianOutput.startCall(String) does not
> > allow to write a header *before* writing the method name, as
> > the protocol requires: a bug ?
>
> Yes. I've just added writeCall and writeMethod, so the
> following would
> work:
>
> writeCall(); // just writes the "c x01 x00" prefix
> writeHeader(String header); // write the header
> writeObject(Object value); // writes the value
> writeMethod(String method); // writes the method name
May I suggest, to be consistent with Burlap, to name the method like this ?
startCall()
In fact, I developed Hessian first, and then I discovered that Burlap does have all the necessary calls :)
This made me wonder if there a specific reason for the fact that there isn't a common interface (say CauchoInput/CauchoOutput) for both Burlap and Hessian ?
The protocol is exactly the same, only its byte representation is different.
I have written such interface in MX4J to factor in a common class the connector implementation, and would have been handy to have it already in hessian/burlap.
> >However I feel my hacks would instead be better placed in an
> improved HessianOutput and HessianServlet than in MX4J code,
> to better support headers. However, I'm not very familiar
> with the uses of Hessian in other contexts (such as the EJB
> world) so I think a detailed discussion with Hessian
> specialists may be the next step, if there is interest.
>
>
> Those two changes should be easy: fixing HessianOutput and
> enhancing the Skeleton/ServiceContext to add headers. Both
> changes fit well into the current architecture.
>
> Both changes are compatible with the EJB support, so there's
> no issues there. They're both just enhancements that other
> uses won't notice.
Cool.
I'm going to use hessian-3.0.8.jar (which also contains the latest Burlap - despite the name - as far as I've understood), it will be easy to move to the improved versions when they will be available.
Do you have an estimate for these fixes ?
Thanks !
Simon
Received on Tue 27 Jul 2004 00:55:28 -0700
This archive was generated by hypermail 2.1.8 : Thu Sep 28 2006 - 20:16:40 PDT