On Mon, 3 Nov 2008 22:10:41 +0800, you wrote:

>>> I do not quite understand this. If the jitter is quantifiable, it would
>>> have a maximum value. So it has determinism. Right? Kind of confusing.
>>
>> Not necessarily. Jitter may be quantified in statistic terms like average,
>> mean, distribution -- without quantifying a maximum value.
>
>You are right. But in the automation segment, often we only care about
>the worst case values and not typical values.

I know, but still... If jitter is not quantified in terms of maximum
value, it may be quantified (in statistical terms) but still doesn't
have a maximum value. Which makes it quantified but not deterministic.

>>> I know there are many applications which do not need hard real time and
>>> allow packet loss (you just wait for the next packet which contains the
>>> same data).
>>
>> There's no form of determinism that can guarantee "no packet loss" -- just
>> cut the cable (or all cables), and there /will/ be packet loss :)
>
>Go back to the topic, are you saying that determinism has nothing
>to do with guaranteeing no data loss. I feel that is a bit strange.

If you mean to say that "network is deterministic" means "network
guarantees no data loss (no matter what)", then I don't think there is
a single deterministic network on this planet, and I doubt there ever
will be. Which would make the term "deterministic" of little value. I
think that in order for it to be useful, the "no matter what" part
needs to be specified differently -- which then means that certain
types of packet loss (like cutting cables) are excluded from the
analysis of whether a network is deterministic or not. 

>> I also think that some authors confuse determinism with reliability.
>So far I have not found any authors do that.

I think the last author you cited did something similar. "Better
determinism" is a quite strange expression. For me, it is or it isn't
deterministic -- this is not a "better determinism" or "worse
determinism".

IMO other characteristics, like reliability, probability of packet
loss, maximum jitter, distribution of jitter etc can be "better" or
"worse". But not determinism.

Gerhard

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist