2008/12/9 Gregg Wonderly <[EMAIL PROTECTED]>:
> Steve Jones wrote:
>>
>>
>> In 2008 its a "bold" statement to say that dynamic discovery and
>> binding isn't an idea you "particularly" buy into.
>>
>> I think most people thought it was bollocks in about 2001. Almost
>> every decent registry product I've seen has been oriented around
>> people usage, its the only way to go. Its a nice reference and they
>> did some nice stuff but it really is what most people have been doing
>> for the last 5+ years.
>
> I believe that Jini lookup has been in wide use since the late 1990s and it
> is
> all about machine based lookup. It's done by type, not by text, so that
> computers can understand the ability of the "service" to perform to the API
> that
> they are looking for. Mobile code allows the implementation to be bound at
> the
> moment of use, not at the moment of development.

But it does require the interfaces to be shared at development and the
registry has to have something that matches the exact interface.  This
is different to the context based lookup approach promoted by UDDI, or
indeed the semantic based lookup that others are these days
presenting.  Back in my more innocent youth I asked Jim Waldo at
JavaOne if he didn't think that an interface based lookup was
restricting and a meta-data based or semantic lookup wouldn't be
better.

His basic response was "good luck with that".

Steve

>
> I know you know this Steve...
>
> Gregg Wonderly
> 

Reply via email to