Pretty clean API. I want to see if Atom Links can replace some of the
published URI schemes so that we can limit the number of URIs exposed by
the system and give more flexibility to the system as a whole. I'm also
wondering if we standardize on Link Relationships rather than data
format, this may free a DTX standard from having to define a data format
altogether.
Also, there's probably is, or going to be support for a compensating
transaction engine as well as we all know, 2pC DTX ain't really that
appropriate for loosly coupled systems.
I have not studied the proposal in depth yet, so I may comment more after I do so. But my immediate response is that I think another less-well-known ...
Bob Haugen
bob.haugen@...
Jun 2, 2009 4:08 pm
I know JBoss TX has a BTP implementation. This here is 2pc dtx restful api. ... -- Bill Burke JBoss, a division of Red Hat http://bill.burkecentral.com...
Hi Bob! It's been a while :-) The compensating transaction model in the reference Bill sent round is what you're looking for I think (cf atoms in BTP). This...
... B2B order-fulfillment scenarios. For example, a quote-to-order sequence is a form of provisional-final 2-phase-commit transaction (with no locking). Does...
Bob Haugen
bob.haugen@...
Jun 7, 2009 2:24 pm
... What if you can't undo, or if undoing is expensive? Then the provisional-final model is better....
Bob Haugen
bob.haugen@...
Jun 7, 2009 2:21 pm
Bob: If you have a model where undo is not an option, Saga won't be of much use. I've run into this very often, but it happens. In those cases, I usually end ...
ha! made a slight slip: "I've run into this very often, but it happens. " should have been: "I've NOT run into this very often, but it happens. " mca ...
... What's wrong with a provisional-final scenario in your opinion? The context is long-running transactions. How long do you want to block?...
Bob Haugen
bob.haugen@...
Jun 7, 2009 2:30 pm
Bob: Well, most often my experience on transactions has been driven by back-end resources that are already committed to 2PC (databases, etc.) so I rarely had...
Yes, but (and this is where we made mistakes in our BTP effort) ONE SIZE DOESN'T FIT ALL USE CASES. Let's make sure we can support multiple protocols if that's...
... I don't think so. ... No comment ;-) ... +1 ... My bad for not checking. We gave a presentation on this last week at JavaOne, but the wiki page doesn't...
... See next exchange of same views as before...but maybe we can make progress this time? ... My opinion remains the same, but maybe I can explain it better...
Bob Haugen
bob.haugen@...
Jun 9, 2009 1:28 pm
... Oh I hope so ;-) ... I think it's the definition of "most" that we will end up debating until the cows come home. If we can agree that there will be a need...
... For the most part, I don't. I think one generic protocol will be most useful, although I can believe that outliers will happen. ... Why does the...
Bob Haugen
bob.haugen@...
Jun 9, 2009 2:22 pm
... If you believe that different signals mean different things even if they have the same name then it does matter. Failure of prepare is different to failure...
If you find yourself in need of a distributed transaction protocol, then how can you possibly say that your architecture is based on REST? I simply cannot see...
... What I have in mind is a coordinator (a client) keeping track of the state of a transaction, and using normal REST interactions with participating...
Bob Haugen
bob.haugen@...
Jun 9, 2009 3:54 pm
... P.S. In this previous message to this group: http://tech.groups.yahoo.com/group/rest-discuss/message/4150 You appeared to be saying something different: ...
Bob Haugen
bob.haugen@...
Jun 9, 2009 4:19 pm
... No, I just found it to be useless for REST. It might still be needed for non-RESTful use of the same protocols. I tried out the above and then simplified...
... This was what I meant in my original post that defining and standardizing a few link relationships might simplify things a lot, both at the client resource...
Instead of looking for a "transaction protocol" over HTTP, consider a resource that is acting as a coordinator. The client can then ask that coordinator...
... I think most would want to design towards a compensation model for transactions (is this what you mean by "saga" Bob?) rather than a 2pc based one even in...
Hello all, During the last year or so we have been working on a RESTful transaction model in the University of Surrey. The goal was to have ACID Transactions, ...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
Offline 
Send Email
