[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Priority revisited: a new primitive

Marcel, Adrian, Ian, Gerald, ...

> I restrict myself to single-execution unit based systems:
> 1. The ordering must be hierarchically composable.
> 2. The hierarchy is the process hierarchy.
> 3. The ordering has to do with the events, not with the processes.
> 4. The processes just specify the order.
> 5. Since processes consist of other processes, they can specify their
>    order by mixing the order of their childeren together with their
>    own influence.
Quick comment - top-of-the-head stuff.

I have trouble with 2. I'm just not convinced there is/should be a process
hierarchy at the definition level (I guess I'm saying that I'd like an event
priority hierarchy, I think). It may be useful in some implementations (e.g.
single execution units) - but if it is then it is to be derived (by a
compiler), not a fundamental description of the system.