Subject: Re: interface gate restriction
From: Rick Reed TSE (rickreed#tseng.co.uk)
Date: Fri Jan 26 2001 - 14:14:41 GMT
anders.ek#telelogic.se at anders.ek#telelogic.se wrote on 25/01/2001 06:46:
> A problem might be that there would be a circular definition. The implicit
> interface of an agent inherits from the interfaces in the incoming signal
> lists. So if we put the implicit interface in the signal list, it would
> inherit from itself.
>
The relevant Z.100 text is 12.1.2 Interface definition, Model:
"The interface defined by a state machine contains in its <interface
specialization> all interfaces given in the incoming signal list associated
with explicit or implicit gates of the state machine. The interface also
contains in its <interface use list> all signals, remote variables and
remote procedures given in the incoming signal list associated with explicit
or implicit gates of the state machine.
The interface defined by an agent or agent type contains in its <interface
specialization> the interface defined by the composite state representing
its state machine.
The interface defined by a type based agent contains in its <interface
specialization> the interface defined by its type."
If the interface is mentioned as an interface in an incoming signal list, it
just "contains" itself in its <interface specialization>.
In 12.1.3 Specialization of data types, Concrete textual grammar:
"The resulting content of a specialized interface definition (<interface
specialization>) consists of the content of the supertypes followed by the
content of the specialized definition. This implies that the set of signals,
remote procedures and remote variables of the specialized interface
definition is the union of those given in the specialized definition itself
and those of the supertypes. The resulting set of definitions must obey the
rules given in 6.3."
So there is just a union of whatever, signals etc. are defined. There does
not seem to be any difficulty resolving the definition.
However, it seems to me that:
If there is neither
a) channel leading the the composite state of an agent creating an implied
gate, nor
b) an explicit gate on the composite state.
the interface of an agent is (by definition) empty. So what happens in the
case of a directly enclosed state machine?
Also, it seems that the interface of an agent does not include any signals
etc. in interfaces of sub-agents of an agent. So there can be gates on the
agent, connected to sub-agents, that do not contribute to the interface of
the agent.
-- Rick Reed - rickreed#tseng.co.uk Tel:+44 1455 55 96 55 Fax:+44 1455 55 96 58 Mob.:+44 7970 50 96 50
This archive was generated by hypermail 2a23 : Mon May 05 2008 - 20:30:55 GMT