Hi Mark,
Well, some things never change ;-). Last I knew 2PC was 2PC was 2PC and has been for 30 years... Sure we have WS-AT now but it is basic 2PC. No heuristics even.
It's true, there's no reason you can't do that. But you can't get guaranteed recovery unless you synchronize the messages with operations on data, meaning holding locks, which is not a great thing to do over the Web.
Eric
----- Original Message ----
From: Mark Baker <[EMAIL PROTECTED]>
To: [email protected]
Sent: Tuesday, July 4, 2006 12:35:52 AM
Subject: Re: [service-orientated-architecture] RESTful lightbulb
From: Mark Baker <[EMAIL PROTECTED]>
To: [email protected]
Sent: Tuesday, July 4, 2006 12:35:52 AM
Subject: Re: [service-orientated-architecture] RESTful lightbulb
On 7/3/06, Eric Newcomer <[EMAIL PROTECTED] com> wrote:
>
>
>
>
> The only thing you don't get here is recovery. This scenario works fine as long as both updates succeed. If one of the updates fails, there's no way to automatically undo the other - at least not with HTTP. This is because HTTP doesn't have persistent sessions, which distributed transaction processing protocols rely upon to share transaction context.
How very 80s of you, Eric. 8-)
Providing a transactional context via a user session is just one way
to do it, and not a particularly good way for today's Internet scale
systems for very many reasons. But there's no reason why a
transactional context can't be provided within the confines of a
stateless messaging architecture.
Mark.
YAHOO! GROUPS LINKS
- Visit your group "service-orientated-architecture" on the web.
- To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]
- Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
