JMS will work with Axis2 as it currently does (that is to say there is a possibility of messages being lost). Reliable JMS will require a little bit more cooperation from AXIS. The specific problem is in reliable, we'd need axis to start a TX when a message is received from a partner (i.e. a request on the server or a response from a server-initiated invoke). I am not sure AXIS is capable of doing this.
As for issues with BART, I'm sure there are some, but I am not aware of them. -mbs On 8/31/07, Alex Boisvert <[EMAIL PROTECTED]> wrote: > > +1. Let's take the trunk (essentially 1.1) and send it into a > maintenance > branch (based on feedback, we might want to do a 1.1.1) > > What are the current options for reliable and transacted invocations with > the two current ILs? For example, does JMS work with Axis2/JBI? How > about > transacted? > > Are there any issues with the BART branch we should be aware of? > > alex > > On 8/31/07, Maciej Szefler <[EMAIL PROTECTED]> wrote: > > > > Hi all, > > > > With 1.1 all spiced and cured, I'd like to merge BART branch into the > > trunk. > > As a a reminder (its been a while) the BART branch is a refactor of the > > engine that provides several benefits: > > > > 1. Simplified integration -- in particular it is no longer necessary to > > start transactions in the IL in order to interface with ODE > > 2. Support for three different "invocation styles": Unreliable (the > > simple, > > no tx no guarantees style), Reliable (transacted transport, useful for > > WS-RM, JMS, and the like), and Transacted (supports atomic invocations). > > 3. Performance -- caching of instance state to eliminate unnecessary > > database reads, reuse of message objects in p2p invocations, elimination > > of > > unnecessary transactions, elimination of unnecessary context switches ( > i.e. > > sequential instance work gets done by the same thread). > > > > Any objects, comments, suggestions are appreciated. > > > > -maciej > > >
