On 7/10/06, Gregg Wonderly <[EMAIL PROTECTED]> wrote:
> Mark, one of the primary issues with your argument is that you suppose that
> when
> there is some common operational theme to methods within an RPC system that
> somehow people would always choose to create a new API so that they would have
>
> enum Bulb {
> HarrysBulb, RitasBulb, FranksBulb, GeorgesBulb
> };
>
> public interface Access100WattLights {
> public void turn100WattBulbOn( Bulb bulb );
> public void turn100WattBulbOff( Bulb bulb );
> }
>
> public interface Access60WattLights {
> public void turn60WattBulbOn( Bulb bulb );
> public void turn60WattBulbOn( Bulb bulb );
> }
I don't assume that. I'm sure that many people would opt to reuse
existing interfaces when given the chance. I just normally use these
kinds of examples to describe how the uniform interface (and therefore
HTTP) is basically the most general, most reusable application
interface (i.e. "invoke" doesn't count 8-) one could conceive of.
> etc. I can only guess that you have some past experience that suggests that
> people who write RPC somehow never use an RPC system/platform/capable language
> which can provide the same level of abstraction that you think only HTTP can
> provide in a RESTful way.
>
> This is where the my arguments are comming from. There's nothing that keeps
> me
> from doing RESTful programming through the RMI programming model. If I do it
> with RMI, then I have transparent access to remote services which might be
> reached via HTTP, JRMP, IIOP, JERI etc. That is the advantage I have. The
> same
> programming environment using an arbitrary transport layer for invocation
> transport/transfer.
It's true, you can use the uniform interface via RMI. But why? A big
part of the advantage of HTTP is that the entire world has already
adopted it for data exchange. Right now, there are something like
O(10^9) software objects out there than can respond to HTTP GET
requests. How many RMI software objects supporting get() are out
there? None. So if you start with RMI, you start from square one.
Now, if the value of being able to use JERI/JRMP/IIOP from RMI is
worth more than that to you, then by all means use it. But I doubt
that's the case for 99.999% of developers.
Mark.
------------------------ Yahoo! Groups Sponsor --------------------~-->
Something is new at Yahoo! Groups. Check out the enhanced email design.
http://us.click.yahoo.com/SISQkA/gOaOAA/yQLSAA/NhFolB/TM
--------------------------------------------------------------------~->
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/service-orientated-architecture/
<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
