I'm moving this to the tuscany-dev list as it looks like it could be
the beginning of a discussion thread.  See comments inline.

  Simon

Thomas Greenwood (JIRA) wrote:
[ https://issues.apache.org/jira/browse/TUSCANY-2035?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12567599#action_12567599 ]
Thomas Greenwood commented on TUSCANY-2035:
-------------------------------------------

My colleagues have created a patch for this, but it's a bit rough and
> has some issues.  We could try and attach a patch file if you think it
> might help.

My apologies for the slow response.  I have been out sick for the last
few days.  I am now back and ready to work on fixing this.  If you can
post the patch soon, I will take a look and use as much of it as I can.

We were thinking that it this change was made then it might be possible
> to start thinking about removing the ThreadMessageContext thread local.
> This would make it easier for people to use threading in Tuscany.

I'd like to discuss this further.  I hadn't been intending to remove
ThreadMessageContext because
 1. It seems like a much more extensive change than is needed to
    fix this problem.
 2. I think the RequestContext API needs something like this to
    provide the correct semantics (but if you have other ideas,
    I'd be very pleased to hear them).
 3. I don't understand why removing this would make it easier for
    people to start using threading in Tuscany, as long as the
    subject problem of this JIRA is fixed.  The only issue I see is
    that the RequestContext API couldn't be used on a non-Tuscany
    thread, and I don't see how removing ThreadMessageContext would
    solve that issue (but if you can give more rationale for why
    this change is helpful, I'd be very interested).

Injected callback references are not resolved at the time of injection
----------------------------------------------------------------------

                Key: TUSCANY-2035
                URL: https://issues.apache.org/jira/browse/TUSCANY-2035
            Project: Tuscany
         Issue Type: Bug
         Components: Java SCA Core Runtime
   Affects Versions: Java-SCA-1.1
        Environment: All
           Reporter: Simon Nash
           Assignee: Simon Nash
            Fix For: Java-SCA-Next


When a callback reference is injected as a proxy or a CallableReference, the 
callback target should be resolved at the time of injection to the caller's 
callback service or the target of any ServiceReference provided by a 
setCallbackObject() call.  Tuscany currently does not do this, but resolves the 
callback target as the time of use, relative to the current invocation context 
at that time.  For example, if A invokes B and C subsequently invokes the same 
B instance, any callbacks from the C invocation through references that were 
injected by the A invocation will go to C instead of going to A as they should.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to