Steven Sharp wrote: >>From: "Arturo Salz" <Arturo.Salz@synopsys.com> > > >>At the recent face2face meeting we discussed and have a preliminary >>agreement that the scheduling should be based on the process call graph >>rather than where the class is defined or the object is instantiated. > > > I was not at the face2face, and I disagree rather strongly with this > idea. A subroutine is an encapsulation of behavior, which should be > independent of where it is called from. Perhaps. We already have exceptions for systf calls in a constant function context as one example. > This idea of using the process call graph also naively assumes pure SV, > and is not easily extensible to mixed-language situations. I have > recently heard of proposals (from SCEMI?) to allow calling exported > DPI tasks from arbitrary places such as SystemC threads. Practical > cross-language calls rely on encapsulation of the behavior of the > subroutine, so they have well-defined behavior no matter where they > are called from. I don't know about others, but I don't think that I was being naive about mixed language. I understand why one would object to the more dynamic scheduling semantics required by the consensus. At the same time, there have been numerous issues involving clocking blocks, program regions, "loop holes" in various attempts to describe this, other semantic irregulariteis, etc, etc. all of which have contributed to the consensus at the face to face that the simplest solution was to accept the more dynamic view of thread-determined behavior. If you are objecting to the consensus, I think that you should also be prepared to have a fairly comprehensive recommendation about how to resolve all of the other issues that have pretty clear semantics with the more dynamic approach. Personally, I would be willing to help refine/comment on such an attempt, but I am not willing to develop one since I believe that the dynamic approach is the most elegant solution that maps the most closely to user expectations and that provides the smallest set of rules to guide user understanding. Gord -- -------------------------------------------------------------------- Gordon Vreugdenhil 503-685-0808 Model Technology (Mentor Graphics) gordonv@model.comReceived on Tue Nov 21 14:38:25 2006
This archive was generated by hypermail 2.1.8 : Tue Nov 21 2006 - 14:39:29 PST