The problem with RMI and Jini was at first marketing, Sun blew it. If
it had been accompanied with a better story from the beginning,
perhaps applications servers would have been Jini-based.
As a stand-alone technology it suffers because of complexity of setup
and configuration. Requiring a registry and lookup service that you
must have running to make remoting work is more weight than most are
willing to deal with. We already have well established IP standards
that provide these services (DNS, DHCP). An ancillary problem is
exception handling. It's simply not robust enough.
We need a remoting library that runs over IP networks that requires
zero configuration, zero setup, like Bonjour. If Bonjour or a Bonjour-
like technology were built into the JVM, RMI would be unnecessary,
and remoting would be cleaner and easier. Although, this does not
address exception handling.
-ernie
On Apr 15, 2006, at 11:15 AM, Dan Creswell wrote:
> Keith Harrison-Broninski wrote:
>> Gervas Douglas wrote:
>>> I have never heard a single simple convincing explanation for why
>>> no tuple
>>> spaces implementation has taken off.
>> I suspect the reason is the same for TSpaces as for Jini. It's
>> just too
>> different, and difficult, for most people to understand.
>>
>> Look how long and hard Gregg has had to work to get people on this
>> list,
>> who must be among the brightest and most techie in IT, to appreciate
>> what Jini has to offer. It's only recently that more than one or two
>> people have started to acknowledge that he may have a point! And
>> unless
>> you have taken the time to actually build code with Jini, you are
>> unlikely to fully "get" it. Taking just one example, witness the
>> recent
>> misunderstanding about how RMI works.
>>
>
> On the subject of misunderstanding RMI - I do a lot of talks,
> coaching,
> mentoring etc in respect of Jini and RMI and have observed the
> following:
>
> Most developers equate RMI to RMI(IIOP) not RMI(JRMP). RMI(JRMP)
> is the
> original RMI with the IIOP based implementation (and all its
> compromises) coming later. This seems to be related in some way to
> the
> widespread use of EJB app servers where RMI(JRMP) was eschewed in
> favour
> of RMI(IIOP).
>
> The result is most developers are under the impression that stubs must
> be generated up front and available on client classpaths in similar
> fashion to the way most will have used CORBA. They aren't aware of
> things such as code downloading or codebases.
>
> Cheers,
>
> Dan.
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
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/
