- From: David Junger <tffy@free.fr>
- Date: Tue, 8 Apr 2014 22:03:29 +0200
- To: Voice List <www-voice@w3.org>
Le 8 apr 2014 à 21:36, Jim Barnett <1jhbarnett@gmail.com> a écrit : > Here's an example that shows why we have to dereference history states when computing the transition domain. Consider a state S with a deep history state H1, plus two child states S1 and S2, which are both deeply nested (have a number of descendent states.) Now add a transition to S1 with target H1 and type 'internal'. What is the domain of this transition? I think that it depends on whether the deep history value contains a descendent of S1 or of S2. If the deep history value is a descendent of S1, then the target of the transition is a descendent of S1, so the transition is internal and the domain is S1. However if the deep history value is a descendent of S2, the transition is not internal and the domain is S. So I think that we do need to dereference the actual history value to compute the domain. Hmmm. Is that what we want, though? Aren't State Charts supposed to be particularly friendly to static analysis, which would be somewhat thrown off by the need to recompute some transitions' LCCA at runtime? not that I'm doing any of that, but some people are. David
Received on Tuesday, 8 April 2014 20:03:59 UTC