W3C home > Mailing lists > Public > www-voice@w3.org > January to March 2007

RE: SCXML: Conflict definition

From: Serge Voloshenyuk <serge_voloshenyuk@yahoo.com>
Date: Wed, 28 Mar 2007 09:51:36 -0700 (PDT)
To: www-voice@w3.org
Message-ID: <861467.91898.qm@web34306.mail.mud.yahoo.com>
Yes, You are right and I'm stupid.
I look at intersection but think about difference.

But now another example.
<scxml xmlns="http://www.w3.org/2005/07/scxml"
       version="1.0"
       initialstate="main">

  <state id="main">
    <initial><transition target="prl"/></initial>
    <transition event="error.*"  target="failure"/>

    <parallel id="prl">
       <state id="p1">
          <transition event="error.*">
             <log expr="_eventdata"/>
          </transition>
       </state>
       <state id="p2">
       </state>
    </parallel>
  </state>

  <final id="failure"/>
  <final id="success"/>
</scxml>
 
According to yours  SelectTransitions algorithm, transition to <failure> from <main> will be executed always.
 Because <p2> branch doesn't have error handling.
 
 But is this behavior what we like? 
 Maybe user have forgotten put error handler just because <p2> doesn't have error prone actions.
 Now he has to put trivial transition to all parallel branches.
 But this can be concerned to other events too (not error.*).
 
 My offer: Searching in the state's ancestors must be stopped on parallel state if at least one branch of the parallel state has matching transition.


Best regards, Serge.

 
---------------------------------
TV dinner still cooling?
Check out "Tonight's Picks" on Yahoo! TV.
Received on Wednesday, 28 March 2007 16:51:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 28 March 2007 16:51:58 GMT