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

Re: Terminology

Dear all,

thanks to Olaf for his fast comments.

first of all, I want to take up Olaf's suggestion. He suggested that
the Petri net type may allow a multiple occurence of the same label 
at the same object.  In fact, there are useful applications of such
a multiple occurence. For instance, if we separate the objects of a
Petri net into groups, a certain place may belong to several groups.
Thus, this place needs more than one label to declare the group(s) to
which that place belongs.  I guess, a Petri net type should explicitly
declare this case of multiple occurence.

Furthermore, Olaf asked some questions and gave some comments about my
proposal for the terminology of the interchange format.  First, I 
summarize these questions and comments.  Second, I will answer these
questions one after the other.  The result will be further questions.


1. Is it allowed to attach labels to labels?
2. May a page contain both net objects and further pages?
3. May an arc cross a page boundary?
4. Are there labels at reference nodes?


ad 1. No, our proposal does not allow labels of labels.  We want to
      keep it simple.  Therefore, we suggest to take the common XML
      techniques for the structuring of a label.

      Is there a usefull application of a label at a label?


ad 2. Yes.


ad 3. No.  The only way to have arcs between objects on different
      pages is via reference nodes.


ad 4. We want to ensure that a net structured in pages may flattened
      in a simple way; in particular without knowing the precise 
      semantics of all labels.  Therefore, we do not allow labels at 
      reference nodes.  Otherwise, if a tool wants to flatten an 
      imported net, the tool has to copy the labels of a reference 
      node to the original node.  The management of several points, 
      where labels at an object are, may be a little bit too tricky.  

      On the other hand, Olaf suggested that a Petri net type may
      allow multiple occurence of a label at the same object.  But, 
      what does it means?  I am not sure, if there is a canonical 
      combining operation for all labels.  The sum of marking 
      labels may be of some sense.  But, there are definitely labels
      without such clear semantics.  For instance duration time:
      the sum or the mean value of all those labels at an place.

      At this point, there are some open questions.  Therefore, we
      suggest not to have labels at reference nodes in the Petri net
      interchange format of level 0.

Kind regards

Michael Weber
-- 
Michael Weber             Tel. +49-30-2093-3075; Fax +49-30-2093-3067
Institut f. Informatik, HU Berlin, Unter den Linden 6, D-10099 Berlin
mailto:mweber@informatik.hu-berlin.de 
http://www2.informatik.hu-berlin.de/~mweber