> We could introduce the streaming call with 'C' x02 x00 instead of 'c'
> x02 x00.
>
> It would probably look like:
>
>    C x02 x02
>    H "header" S ...
>    m "method name"
>    data_1
>    data_2
>    ...
>    data_n
>    z

So this means you find out whether it's streaming after invoking
input.readCall()?
Therefore, based on the return code, you would then know how to
materialize the rest of the input or tokenize the stream and invoke an
asynchronous handler.

Did you mean C x02 x02 or C x02 x00 (or is this just detail that
doesn't matter yet)?

>
> I'd need to look at BEEP again, it's been years since I've looked at
> it and I can't remember much about it.

I've recently revisited it but it seems not to have caught on, as many
of the old implementations are now effectively dead. I like the simple
remit of Hessian, and adding too many features into it would be
detremental - leaving it open for extension by different layers
improves the seperation of concerns.

Having said that, I imagine you could introduce a standardized scheme
to acheive some of the goals of BEEP by using Hessian headers if you
wanted to. By doing that, you wouldn't bloat the underlying encoding
specification, you would only be introducing standardized message
interaction primitives like encryption, authentication, multiplexing,
pipelining, keeping connections alive etc.

>
> That's a good point.  I think the cleanest solution would be to reset
> the references after each top-level object, i.e. start counting from
> zero.  Anything else would be tricky to specify.

I agree.

-Ben


_______________________________________________
hessian-interest mailing list
[email protected]
http://maillist.caucho.com/mailman/listinfo/hessian-interest

Reply via email to