MSC-News: Partial and Total ordering of MSC


Subject: MSC-News: Partial and Total ordering of MSC
From: Aldo Buratti (buraldo#maganet.net)
Date: Wed Nov 04 1998 - 23:02:35 GMT


I have a clarification request about the partial ordering of MSC events.

Let's talk of the MSC "event_ordering" taken from Z.120 (ver 10/96),
chapter 4, Fig.4.1a
(You can look at the attached GIF copy named
"event_ordering_original.gif)
[NOTE: the frame of the MSC has been replaced with a fictitious axis
named "ENV"]

As you can read on the Z.120 document, just above the figure, the
following partial relations are listed:
1) out(m2) < in(m2)
2) out(m3) < in(m3)
3) out(m4) < in(m4)
4) in(m1) < out(m2) < out(m3) < in(m4)
5) in(m2) < out(m4)

Now, we can start the discussion:
i) relations 1,2,3 are the trivial relations of each message, BUT, WHY
the relation "out(m1) < in(m1)" has not been listed ?
ii) relation 4 ...sorry... I don't understand !!
  In my opinion it would be:
        out(m1) < out(m2) < out(m3) [total order of axis ENV],
  PLUS in(m1) < in(m4) [total order of axis A]

I'm not sure of the intended meaning of the "<" relation, but I argue
that the original partial order relations are inconsistent with the
original picture.
Maybe, the correct picture would be the second attached picture named
"event_ordering_NEW".

What's right and what's wrong ?

                    ==============================
A second question:
------------------
In chapter 2.4, it is stated that:
1) "If an incoming event and an outgoing event are on the same point of
an instance axis, then it is interpreted as if the incoming event is
drawn above the outgoing event"
...
2) "It is not allowed to draw two or more outgoing events on the same
point."
3) "It is not allowed to draw two or more incoming events on the same
point."
..

I agree with (1) (it's a useful shorthand) , but what's the reason of
the rules (2) and (3) ??
In my opinion these are unnecessary restrictions (and in my experience I
saw a lot of such patterns).

Don't you think that these widely used patterns (even in an informal
way), can be legitimated by adopting an "implicit" coregion ?
In other words, two incoming/outcoming events on the same point, should
be treated as events occurring at "the same time", that is in a random
order, that is like in a coregion.

 I hope to hear your opinions, soon

                                     Sincerely
                                    Aldo Buratti
------------------------------------------------------------
 Aldo Buratti
 Irrational Numbers
 Milano - Italy
------------------------------------------------------------





This archive was generated by hypermail 2a23 : Wed Jun 19 2013 - 13:16:38 GMT