Forums

I am okay with stefan's modifications to the proposal. It is simple
to warn the user when events are firing simultaneously without
priorities. If one wants truely random without a nondeterminism
package just give all an equal priority.

Chris

Sent from my iPod

On Jun 25, 2010, at 10:12 AM, Stefan Hoops <shoops@vbi.vt.edu> wrote:

> Hello Lucian,
>
> On Fri, 25 Jun 2010 16:36:39 +0100
> Lucian Smith <lpsmith@spod-central.org> wrote:
>
>> * shoops <shoops@vbi.vt.edu> [2010-06-25 14:14] writes:
>>> Hello Lucian,
>>>
>>> On Fri, 25 Jun 2010 07:06:24 +0100
>>> Lucian Smith <lpsmith@spod-central.org> wrote:
>>>
>>>>> 3) Some simultaneously executed events have the priority object
>>>>> and some not.
>>>>> Not defined as in L2V4 and L3 currently
>>>>
>>>> Yes, except that in #3, those events with priorities must execute
>>>> in priority order. It's only the events without priorities that
>>>> may execute in any order.
>>>
>>> This is not possible. You have a set of elements of which some can
>>> be ordered and some not. This leads to the situation that the set
>>> cannot be ordered, i.e., the behavior is undefined. Every other
>>> interpretation would have to introduce some defaults which are
>>> strictly prohibited. Note that executing the events in a random
>>> order or partially random order as yu usggest is different than the
>>> order being undefined. People can easily avoid this situation by
>>> specifying the priority attribute for all events.
>>
>> I think you're misunderstanding me. If you have six events, three
>> with priorities and three without:
>>
>> 10 7 2 X X X
>>
>> The X's can be inserted anywhere, *but* the 10 must fire before the 7
>> which must fire before the 2.
>>
>> X X X 10 7 2
>> X 10 X X 7 2
>> 10 X 7 X 2 X
>> 10 7 2 X X X
>>
>> but not:
>>
>> 7 2 X 10 X X
>>
>> There are no defaults, and the overall behavior is undefined... but
>> the priority-flagged events still fire in a defined order with
>> respect to each other.
>
> No you are misunderstanding me! I was afraid that the above was your
> intend. What you propose is that we change the undefined behavior to
> random insertions into a sorted set. I fact you are saying that the
> default for the optional object priority is a random number if and
> only
> if it happens simultaneously with events that have a priority object.
> You however do not specify from which distribution the random number
> shall be drawn. We should avoid all this mess and simply leave this as
> undefined. If people want the above they need to define the proper
> priorities for all events. This proper definition could include a
> distribution package as suggested by Darren if you want a random
> order.
>
> Please avoid shortcuts and overloaded semantics at all costs. We made
> such an effort to become very clear and specific in L3. We should not
> haste the introduction of new features. This does not mean that I am
> against the priority object in fact I am very much in favor of it.
> However I can not stress enough that we should make its semantics as
> simple as possible.
>
> Thanks,
> Stefan
>
>
>>
>> -Lucian
>> ____________________________________________________________
>> To manage your sbml-discuss list subscription, visit
>> https://utils.its.caltech.edu/mailman/listinfo/sbml-discuss
>>
>> For a web interface to the sbml-discuss mailing list, visit
>> http://sbml.org/Forums/
>>
>> For questions or feedback about the sbml-discuss list,
>> contact sbml-team@caltech.edu
>
>
> --
> Stefan Hoops, Ph.D.
> Senior Project Associate
> Virginia Bioinformatics Institute - 0477
> Virginia Tech
> Bioinformatics Facility II
> Blacksburg, Va 24061, USA
>
> Phone: (540) 231-1799
> Fax: (540) 231-2606
> Email: shoops@vbi.vt.edu
> ____________________________________________________________
> To manage your sbml-discuss list subscription, visit
> https://utils.its.caltech.edu/mailman/listinfo/sbml-discuss
>
> For a web interface to the sbml-discuss mailing list, visit
> http://sbml.org/Forums/
>
> For questions or feedback about the sbml-discuss list,
> contact sbml-team@caltech.edu
____________________________________________________________
To manage your sbml-discuss list subscription, visit
https://utils.its.caltech.edu/mailman/listinfo/sbml-discuss

For a web interface to the sbml-discuss mailing list, visit
http://sbml.org/Forums/

For questions or feedback about the sbml-discuss list,
contact sbml-team@caltech.edu