Eric Newcomer wrote:
> I just wanted to point out the common confusion
> between a "request" for a transaction and the
> execution of the transaction itself. Mobile devices
> are capable of the former today but not the latter.
>
> But as I said I agree with you that this would be a
> good thing, and I also believe it's achievable in the
> not too distant future.
You may feel like a mobile device can't participate in a transaction, literaly,
but that is simply not true.
A simple transaction participant using the Jini transaction model, can exist in
a cell phone, and perform all tasks associated with the Jini 2 phase commit
model.
Other transaction systems may not be as "open" to transport and invocation
plugability as the JERI layer is. But, certainly, you can implement this,
quite
easily.
The transaction voting will require the device to be visible to the transaction
manager. But, that is an endpoint implementation issue. The connection from
the transaction manager, out to the mobile device, can be managed as a
connection from the mobile device back into the service. At that point the
vote
request can go out to the mobile device, and it can send back its vote then, or
on a subsequent connection.
The Jini Transaction timeout and leasing mechanisms allow you to set upper
limits on when the transaction voting action should timeout (if ever) and how
long the device can be disconnected before the transaction participant endpoint
lease will expire and cause the transaction to be cancelled.
The article the Gervas point doesn't detail at this level, what they are doing
with that particular development. But, these guys are experts with the Jini
software model, and they have stuff working last I heard.
Gregg Wonderly
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/
